TADPOLE_BSWiMS

1 TADPOLE and BSWiMS

1.0.1 Loading the libraries

library("FRESA.CAD")
library(readxl)
library(igraph)
op <- par(no.readonly = TRUE)
pander::panderOptions('digits', 3)
pander::panderOptions('table.split.table', 400)
pander::panderOptions('keep.trailing.zeros',TRUE)

1.1 The data set

TADPOLE_D1_D2 <- read.csv("~/GitHub/BSWiMS/Data/TADPOLE/TADPOLE_D1_D2.csv")
TADPOLE_D1_D2_Dict <- read.csv("~/GitHub/BSWiMS/Data/TADPOLE/TADPOLE_D1_D2_Dict.csv")
TADPOLE_D1_D2_Dict_LR <- as.data.frame(read_excel("~/GitHub/BSWiMS/Data/TADPOLE/TADPOLE_D1_D2_Dict_LR.xlsx",sheet = "LeftRightFeatures"))


rownames(TADPOLE_D1_D2_Dict) <- TADPOLE_D1_D2_Dict$FLDNAME

1.2 Conditioning the data

# mm3 to mm
isVolume <- c("Ventricles","Hippocampus","WholeBrain","Entorhinal","Fusiform","MidTemp","ICV",
              TADPOLE_D1_D2_Dict$FLDNAME[str_detect(TADPOLE_D1_D2_Dict$TEXT,"Volume")]
              )


#TADPOLE_D1_D2[,isVolume] <- apply(TADPOLE_D1_D2[,isVolume],2,'^',(1/3))
TADPOLE_D1_D2[,isVolume] <- TADPOLE_D1_D2[,isVolume]^(1/3)

# mm2 to mm
isArea <- TADPOLE_D1_D2_Dict$FLDNAME[str_detect(TADPOLE_D1_D2_Dict$TEXT,"Area")]
TADPOLE_D1_D2[,isArea] <- sqrt(TADPOLE_D1_D2[,isArea])

# Get only cross sectional measurements
FreeSurfersetCross <- str_detect(colnames(TADPOLE_D1_D2),"UCSFFSX")

# The subset of baseline measurements
baselineTadpole <- subset(TADPOLE_D1_D2,VISCODE=="bl")
table(baselineTadpole$DX)

                   Dementia Dementia to MCI             MCI MCI to Dementia 
          7             336               1             864               5 
  MCI to NL              NL       NL to MCI 
          2             521               1 

rownames(baselineTadpole) <- baselineTadpole$PTID


validBaselineTadpole <- cbind(DX=baselineTadpole$DX,
                                 AGE=baselineTadpole$AGE,
                                 Gender=1*(baselineTadpole$PTGENDER=="Female"),
                                 ADAS11=baselineTadpole$ADAS11,
                                 ADAS13=baselineTadpole$ADAS13,
                                 MMSE=baselineTadpole$MMSE,
                                 RAVLT_immediate=baselineTadpole$RAVLT_immediate,
                                 RAVLT_learning=baselineTadpole$RAVLT_learning,
                                 RAVLT_forgetting=baselineTadpole$RAVLT_forgetting,
                                 RAVLT_perc_forgetting=baselineTadpole$RAVLT_perc_forgetting,
                                 FAQ=baselineTadpole$FAQ,
                                 Ventricles=baselineTadpole$Ventricles,
                                 Hippocampus=baselineTadpole$Hippocampus,
                                 WholeBrain=baselineTadpole$WholeBrain,
                                 Entorhinal=baselineTadpole$Entorhinal,
                                 Fusiform=baselineTadpole$Fusiform,
                                 MidTemp=baselineTadpole$MidTemp,
                                 ICV=baselineTadpole$ICV,
                                 baselineTadpole[,FreeSurfersetCross])


LeftFields <- TADPOLE_D1_D2_Dict_LR$LFN
names(LeftFields) <- LeftFields
LeftFields <- LeftFields[LeftFields %in% colnames(validBaselineTadpole)]
RightFields <- TADPOLE_D1_D2_Dict_LR$RFN
names(RightFields) <- RightFields
RightFields <- RightFields[RightFields %in% colnames(validBaselineTadpole)]

## Normalize to ICV
validBaselineTadpole$Ventricles=validBaselineTadpole$Ventricles/validBaselineTadpole$ICV
validBaselineTadpole$Hippocampus=validBaselineTadpole$Hippocampus/validBaselineTadpole$ICV
validBaselineTadpole$WholeBrain=validBaselineTadpole$WholeBrain/validBaselineTadpole$ICV
validBaselineTadpole$Entorhinal=validBaselineTadpole$Entorhinal/validBaselineTadpole$ICV
validBaselineTadpole$Fusiform=validBaselineTadpole$Fusiform/validBaselineTadpole$ICV
validBaselineTadpole$MidTemp=validBaselineTadpole$MidTemp/validBaselineTadpole$ICV

leftData <- validBaselineTadpole[,LeftFields]/validBaselineTadpole$ICV
RightData <- validBaselineTadpole[,RightFields]/validBaselineTadpole$ICV

## get mean and relative difference 
meanLeftRight <- (leftData + RightData)/2
difLeftRight <- abs(leftData - RightData)
reldifLeftRight <- difLeftRight/meanLeftRight
colnames(meanLeftRight) <- paste("M",colnames(meanLeftRight),sep="_")
colnames(difLeftRight) <- paste("D",colnames(difLeftRight),sep="_")
colnames(reldifLeftRight) <- paste("RD",colnames(reldifLeftRight),sep="_")


validBaselineTadpole <- validBaselineTadpole[,!(colnames(validBaselineTadpole) %in% 
                                               c(LeftFields,RightFields))]
validBaselineTadpole <- cbind(validBaselineTadpole,meanLeftRight,reldifLeftRight)

## Remove columns with too many NA more than %15 of NA
nacount <- apply(is.na(validBaselineTadpole),2,sum)/nrow(validBaselineTadpole) < 0.15
diagnose <- validBaselineTadpole$DX
pander::pander(table(diagnose))

	Dementia	Dementia to MCI	MCI	MCI to Dementia	MCI to NL	NL	NL to MCI
7	336	1	864	5	2	521	1

validBaselineTadpole <- validBaselineTadpole[,nacount]
## Remove character columns
ischar <- sapply(validBaselineTadpole,class) == "character"
validBaselineTadpole <- validBaselineTadpole[,!ischar]
## Place back diagnose
validBaselineTadpole$DX <- diagnose


validBaselineTadpole <- validBaselineTadpole[complete.cases(validBaselineTadpole),]
ischar <- sapply(validBaselineTadpole,class) == "character"
validBaselineTadpole[,!ischar] <- sapply(validBaselineTadpole[,!ischar],as.numeric)

colnames(validBaselineTadpole) <- str_remove_all(colnames(validBaselineTadpole),"_UCSFFSX_11_02_15_UCSFFSX51_08_01_16")
colnames(validBaselineTadpole) <- str_replace_all(colnames(validBaselineTadpole)," ","_")
validBaselineTadpole$LONISID <- NULL
validBaselineTadpole$IMAGEUID <- NULL
validBaselineTadpole$LONIUID <- NULL

diagnose <- as.character(validBaselineTadpole$DX)
validBaselineTadpole$DX <- diagnose
pander::pander(table(validBaselineTadpole$DX))

Dementia	Dementia to MCI	MCI	MCI to Dementia	MCI to NL	NL	NL to MCI
244	1	711	2	2	452	1

validDX <- c("NL","MCI","Dementia")

validBaselineTadpole <- validBaselineTadpole[validBaselineTadpole$DX %in% validDX,]
validBaselineTadpole$DX <- as.factor(validBaselineTadpole$DX)
pander::pander(table(validBaselineTadpole$DX))

Dementia	MCI	NL
244	711	452

1.3 Get the Time To Event on MCI Subjects

subjectsID <- rownames(validBaselineTadpole)
visitsID <- unique(TADPOLE_D1_D2$VISCODE)
baseDx <- TADPOLE_D1_D2[TADPOLE_D1_D2$VISCODE=="bl",c("PTID","DX","EXAMDATE")]
rownames(baseDx) <- baseDx$PTID 
baseDx <- baseDx[subjectsID,]
lastDx <- baseDx
toDementia <- baseDx
table(lastDx$DX)

Dementia MCI NL 244 711 452

hasDementia <- lastDx$PTID[str_detect(lastDx$DX,"Dementia")]


for (vid in visitsID)
{
  DxValue <- TADPOLE_D1_D2[TADPOLE_D1_D2$VISCODE==vid,c("PTID","DX","EXAMDATE")]
  rownames(DxValue) <- DxValue$PTID 
  DxValue <- DxValue[DxValue$PTID %in% subjectsID,]
  noDX <- DxValue$PTID[nchar(DxValue$DX) < 1]
  print(length(noDX))
  DxValue[noDX,] <- lastDx[noDX,]
  inLast <- lastDx$PTID[lastDx$PTID %in% DxValue$PTID]
  print(length(inLast))
  lastDx[inLast,] <- DxValue[inLast,]
  noDementia <- !(toDementia$PTID %in% hasDementia)
  toDementia[noDementia,] <- lastDx[noDementia,]
  hasDementia <- unique(c(hasDementia,lastDx$PTID[str_detect(lastDx$DX,"Dementia")]))
}

[1] 0 [1] 1407 [1] 2 [1] 1320 [1] 6 [1] 1212 [1] 23 [1] 1090 [1] 802 [1] 1054 [1] 29 [1] 706 [1] 20 [1] 212 [1] 14 [1] 167 [1] 32 [1] 551 [1] 25 [1] 297 [1] 18 [1] 130 [1] 665 [1] 665 [1] 112 [1] 112 [1] 176 [1] 176 [1] 177 [1] 177 [1] 624 [1] 624 [1] 251 [1] 251 [1] 159 [1] 159 [1] 7 [1] 7 [1] 17 [1] 99 [1] 9 [1] 63 [1] 1 [1] 1

table(lastDx$DX)

   Dementia Dementia to MCI             MCI MCI to Dementia       MCI to NL 
        426               2             460              80               7 
         NL  NL to Dementia       NL to MCI 
        405               1              26 

baseMCI <-baseDx$PTID[baseDx$DX == "MCI"]
lastDementia <- lastDx$PTID[str_detect(lastDx$DX,"Dementia")]
lastDementia2 <- toDementia$PTID[str_detect(toDementia$DX,"Dementia")]
lastNL <- lastDx$PTID[str_detect(lastDx$DX,"NL")]

MCIatBaseline <- baseDx[baseMCI,]
MCIatEvent <- toDementia[baseMCI,]
MCIatLast <- lastDx[baseMCI,]

MCIconverters <- MCIatBaseline[baseMCI %in% lastDementia,]
MCI_No_converters <- MCIatBaseline[!(baseMCI %in% MCIconverters$PTID),]
MCIconverters$TimeToEvent <- (as.Date(toDementia[MCIconverters$PTID,"EXAMDATE"]) 
                                   - as.Date(MCIconverters$EXAMDATE))

sum(MCIconverters$TimeToEvent ==0)

[1] 0

MCIconverters$AtEventDX <- MCIatEvent[MCIconverters$PTID,"DX"]
MCIconverters$LastDX <- MCIatLast[MCIconverters$PTID,"DX"]

MCI_No_converters$TimeToEvent <- (as.Date(lastDx[MCI_No_converters$PTID,"EXAMDATE"]) 
                                   - as.Date(MCI_No_converters$EXAMDATE))

MCI_No_converters$LastDX <- MCIatLast[MCI_No_converters$PTID,"DX"]

MCI_No_converters <- subset(MCI_No_converters,TimeToEvent > 0)

1.3.1 Correlation Matrix Data

The heat map of the testing set.

cormat <- cor(validBaselineTadpole[,colnames(validBaselineTadpole) != "DX"],method="spearman")
diag(cormat) <- 0;
corrmax <- apply(cormat,2,max)
whomax <- colnames(cormat)[corrmax>0.75]
gplots::heatmap.2(abs(cormat[whomax,whomax]),
                  trace = "none",
                  scale = "none",
                  mar = c(10,10),
                  col=rev(heat.colors(5)),
                  main = "Baseline TADPOLE Correlation",
                  cexRow = 0.5,
                  cexCol = 0.5,
                  key.title=NA,
                  key.xlab="Spearman Correlation",
                  xlab="Feature", ylab="Feature")

par(op)

2 Predicting ADAS13

Here we will diagnose ADAS13

2.1 Training and testing sets

TrainFraction <- 0.5;

TADPOLECrossMRI <- validBaselineTadpole
summary(TADPOLECrossMRI$ADAS13)

Min. 1st Qu. Median Mean 3rd Qu. Max. 0.00 9.00 14.67 16.37 22.00 51.00

TADPOLECrossMRI$ADAS11 <- NULL
TADPOLECrossMRI$MMSE <- NULL
TADPOLECrossMRI$RAVLT_immediate <- NULL
TADPOLECrossMRI$RAVLT_learning <- NULL
TADPOLECrossMRI$RAVLT_perc_forgetting <- NULL
TADPOLECrossMRI$RAVLT_forgetting <- NULL
TADPOLECrossMRI$FAQ <- NULL
TADPOLECrossMRI$DX <- NULL

trainSet <- sample(nrow(TADPOLECrossMRI),nrow(TADPOLECrossMRI)*TrainFraction)

TADPOLECrossMRITrain <- TADPOLECrossMRI[trainSet,]
TADPOLECrossMRITest <- TADPOLECrossMRI[-trainSet,]

2.1.1 Learning ADAS13

bml <- BSWiMS.model(ADAS13~.,TADPOLECrossMRITrain,maxTrainModelSize=50,NumberofRepeats = 20)

[++-+-++-++-++-++-+-+-+-+-++-+-+-+-+-+–+–+-+-++-]..

plot(bml$bagging$avgLogPvalues,bml$bagging$bagged.model$baggingAnalysis$dMSE_values)

#bml$bagging$bagged.model$wts.List
#bml$bagging$bagged.model$baggingAnalysis$wts

pander::pander(bml$bagging$Jaccard.SM)

0.366

fs <- bml$bagging$frequencyTable
barplot(fs[order(-fs)],las=2,main="Selected Features",cex.names = 0.5)

sm <- summary(bml)
sm$MSE

[1] 44.37987

pander::pander(sm$coefficients)

	Estimate	lower	mean	upper	u.MSE	r.MSE	model.MSE	NeRI	F.pvalue	t.pvalue	Sign.pvalue	Wilcox.pvalue	%VarExp	Frequency
Hippocampus	-149.7432	-210.157	-149.7432	-8.93e+01	66.5	50.7	47.8	0.05932	0.00e+00	6.25e-05	0.04588	5.01e-03	0.03056	1.00
M_ST24CV	-283.8614	-369.063	-283.8614	-1.99e+02	68.5	54.7	48.9	0.10295	0.00e+00	1.29e-08	0.00174	6.69e-05	0.06017	1.00
M_ST31TA	-358.9971	-513.816	-358.9971	-2.04e+02	80.9	54.4	50.5	0.05349	1.31e-13	7.66e-05	0.07886	9.00e-03	0.04098	0.10
M_ST24TA	-264.3159	-421.053	-264.3159	-1.08e+02	63.9	49.1	46.9	0.06421	9.06e-09	6.21e-04	0.04245	1.54e-02	0.02240	0.95
M_ST60CV	85.1582	27.387	85.1582	1.43e+02	88.6	48.5	46.8	0.00977	5.42e-07	4.00e-02	0.37029	1.66e-01	0.01670	0.75
M_ST51TA	60.6724	19.158	60.6724	1.02e+02	92.3	47.5	46.0	0.00618	6.23e-07	3.69e-01	0.53640	4.67e-01	0.01608	0.10
M_ST40TA	-350.0541	-597.735	-350.0541	-1.02e+02	73.8	52.2	50.6	0.05602	1.14e-06	3.31e-02	0.06814	4.61e-02	0.01684	0.25
M_ST30SV	52.4153	15.133	52.4153	8.97e+01	71.5	48.4	46.9	0.08155	1.24e-06	3.25e-03	0.01378	1.58e-02	0.01547	1.00
M_ST24SA	31.6447	6.909	31.6447	5.64e+01	95.2	51.4	49.8	0.07605	1.39e-06	4.16e-05	0.02171	3.05e-03	0.01641	0.25
M_ST51CV	20.7792	4.162	20.7792	3.74e+01	93.5	51.5	50.1	0.03220	5.42e-06	7.61e-02	0.20469	1.78e-01	0.01438	0.10
M_ST56TA	-89.7854	-160.511	-89.7854	-1.91e+01	86.7	46.9	45.7	0.01291	9.33e-06	1.03e-01	0.29347	1.94e-01	0.01240	0.15
RD_ST31TA	13.3925	2.637	13.3925	2.41e+01	89.3	48.1	46.9	0.06955	1.12e-05	7.31e-03	0.03309	6.14e-02	0.01250	0.95
AGE	-0.0676	-0.123	-0.0676	-1.23e-02	95.5	48.0	46.8	0.02879	1.68e-05	1.86e-03	0.21585	5.24e-02	0.01195	0.75
M_ST31CV	-31.1982	-56.297	-31.1982	-6.10e+00	78.7	50.5	49.3	0.04933	1.96e-05	7.64e-03	0.09068	5.13e-02	0.01230	0.30
M_ST40SA	-13.7941	-25.402	-13.7941	-2.19e+00	87.1	52.0	50.8	0.03356	2.40e-05	1.76e-02	0.19417	6.40e-02	0.01243	0.15
M_ST12SV	-29.2047	-55.363	-29.2047	-3.05e+00	71.6	52.3	51.1	0.03954	2.94e-05	3.55e-02	0.15555	7.31e-02	0.01222	0.10
M_ST40CV	-75.5757	-140.070	-75.5757	-1.11e+01	70.8	48.0	46.9	0.04585	3.34e-05	1.98e-02	0.10197	6.54e-02	0.01116	1.00
M_ST129TS	734.2100	63.601	734.2100	1.40e+03	95.1	49.6	48.5	0.03896	4.22e-05	6.67e-03	0.15425	3.41e-02	0.01113	0.75
ST2SV	-0.3445	-0.673	-0.3445	-1.57e-02	92.8	51.5	50.5	0.03435	6.87e-05	6.10e-02	0.17434	1.20e-01	0.01084	0.15
M_ST62TA	176.8440	10.859	176.8440	3.43e+02	94.5	50.7	49.7	0.02988	7.82e-05	1.06e-01	0.21608	1.54e-01	0.01064	0.30
RD_ST40CV	12.6473	0.692	12.6473	2.46e+01	90.6	47.7	46.8	0.05949	1.03e-04	1.07e-02	0.05615	4.56e-02	0.00954	0.90
RD_ST59SA	-12.2523	-23.944	-12.2523	-5.61e-01	97.0	48.8	47.9	0.01841	1.07e-04	8.41e-02	0.32911	2.06e-01	0.00972	0.80
RD_ST129TA	2.5023	0.119	2.5023	4.89e+00	93.0	50.9	50.0	0.04694	1.20e-04	6.06e-02	0.11372	1.45e-01	0.00996	0.05
M_ST56TS	342.4211	-9.562	342.4211	6.94e+02	95.1	50.5	49.6	0.02067	1.29e-04	6.52e-02	0.27752	1.70e-01	0.00984	0.20
M_ST49TS	45.0263	1.743	45.0263	8.83e+01	96.3	48.7	47.7	0.08962	1.33e-04	3.26e-03	0.00965	2.36e-02	0.00938	0.05
M_ST60SA	38.0636	4.099	38.0636	7.20e+01	96.0	50.9	49.9	0.03652	1.45e-04	3.05e-02	0.15264	1.06e-01	0.00971	0.45
M_ST35TA	191.2065	-0.476	191.2065	3.83e+02	89.9	51.4	50.5	0.03824	2.05e-04	6.23e-02	0.16414	1.96e-01	0.00929	0.10
M_ST61SV	41.9162	-2.011	41.9162	8.58e+01	95.4	51.0	50.1	0.05985	2.19e-04	3.06e-02	0.05984	5.35e-02	0.00916	0.15
M_ST55CV	-10.0714	-20.277	-10.0714	1.34e-01	85.4	47.6	46.8	0.04711	2.34e-04	3.45e-02	0.11165	8.86e-02	0.00845	0.15
RD_ST14TS	-0.3893	-0.789	-0.3893	9.96e-03	95.6	48.1	47.3	0.00595	2.63e-04	3.29e-02	0.45199	2.62e-01	0.00838	0.10
M_ST55SA	-6.5188	-13.185	-6.5188	1.47e-01	94.5	50.1	49.3	0.05255	2.63e-04	1.63e-02	0.07364	4.46e-02	0.00875	0.15
RD_ST52TA	5.5492	-0.698	5.5492	1.18e+01	93.2	50.8	50.0	0.02550	2.64e-04	3.92e-02	0.26146	1.30e-01	0.00891	0.15
M_ST15CV	-11.3999	-23.523	-11.3999	7.23e-01	88.6	50.7	49.9	0.06686	3.79e-04	1.08e-02	0.04134	5.71e-02	0.00835	0.05
M_ST32CV	-21.2796	-44.120	-21.2796	1.56e+00	74.9	51.2	50.4	0.03964	4.00e-04	1.64e-02	0.15463	6.28e-02	0.00835	0.10
M_ST60TS	146.9471	-10.105	146.9471	3.04e+02	93.0	51.2	50.4	0.00748	4.04e-04	4.45e-02	0.58095	1.75e-01	0.00835	0.10
RD_ST40TA	5.7991	-0.528	5.7991	1.21e+01	87.0	51.5	50.7	0.08337	4.58e-04	9.26e-03	0.01102	1.50e-02	0.00821	0.20
Ventricles	-6.9694	-14.695	-6.9694	7.57e-01	86.3	47.3	46.5	0.04350	5.03e-04	3.12e-03	0.12402	2.91e-02	0.00743	0.45
RD_ST57TA	14.1947	-1.982	14.1947	3.04e+01	91.7	50.6	49.9	0.05413	6.32e-04	4.50e-02	0.07549	4.46e-02	0.00766	0.40
RD_ST24TA	1.3081	-0.688	1.3081	3.30e+00	89.5	50.9	50.4	0.02814	4.92e-03	2.18e-01	0.23612	2.35e-01	0.00494	0.10
M_ST23CV	17.9270	-9.340	17.9270	4.52e+01	95.4	51.5	51.0	0.04450	5.07e-03	3.13e-01	0.11307	2.71e-01	0.00505	0.10
RD_ST56TA	12.0381	-7.646	12.0381	3.17e+01	94.4	50.4	50.0	-0.03626	6.50e-03	5.37e-01	1.00000	5.83e-01	0.00456	0.40
WholeBrain	-3.7858	-11.958	-3.7858	4.39e+00	80.1	50.2	49.9	0.00711	1.98e-02	5.13e-01	0.44005	5.67e-01	0.00310	0.05

pander::pander(bml$univariate[bml$selectedfeatures,])

	Name	RName	ZUni
Ventricles	Ventricles	Ventricles	Inf
M_ST24TA	M_ST24TA	M_ST24TA	Inf
M_ST40CV	M_ST40CV	M_ST40CV	Inf
M_ST30SV	M_ST30SV	M_ST30SV	Inf
Hippocampus	Hippocampus	Hippocampus	Inf
RD_ST31TA	RD_ST31TA	RD_ST31TA	7.71
M_ST24CV	M_ST24CV	M_ST24CV	Inf
RD_ST40CV	RD_ST40CV	RD_ST40CV	7.01
RD_ST59SA	RD_ST59SA	RD_ST59SA	1.15
M_ST60CV	M_ST60CV	M_ST60CV	8.04
M_ST55CV	M_ST55CV	M_ST55CV	Inf
M_ST51TA	M_ST51TA	M_ST51TA	6.00
AGE	AGE	AGE	3.47
M_ST129TS	M_ST129TS	M_ST129TS	3.87
M_ST49TS	M_ST49TS	M_ST49TS	2.53
RD_ST14TS	RD_ST14TS	RD_ST14TS	3.38
M_ST56TA	M_ST56TA	M_ST56TA	Inf
RD_ST56TA	RD_ST56TA	RD_ST56TA	4.51
M_ST62TA	M_ST62TA	M_ST62TA	4.40
M_ST31CV	M_ST31CV	M_ST31CV	Inf
WholeBrain	WholeBrain	WholeBrain	Inf
RD_ST24TA	RD_ST24TA	RD_ST24TA	7.57
M_ST24SA	M_ST24SA	M_ST24SA	3.74
M_ST60SA	M_ST60SA	M_ST60SA	2.84
M_ST56TS	M_ST56TS	M_ST56TS	3.86
M_ST40TA	M_ST40TA	M_ST40TA	Inf
RD_ST57TA	RD_ST57TA	RD_ST57TA	6.36
RD_ST52TA	RD_ST52TA	RD_ST52TA	5.41
M_ST12SV	M_ST12SV	M_ST12SV	Inf
M_ST32CV	M_ST32CV	M_ST32CV	Inf
M_ST60TS	M_ST60TS	M_ST60TS	5.52
M_ST61SV	M_ST61SV	M_ST61SV	3.60
M_ST51CV	M_ST51CV	M_ST51CV	5.21
M_ST23CV	M_ST23CV	M_ST23CV	3.58
M_ST31TA	M_ST31TA	M_ST31TA	Inf
ST2SV	ST2SV	ST2SV	5.69
M_ST55SA	M_ST55SA	M_ST55SA	4.38
RD_ST40TA	RD_ST40TA	RD_ST40TA	Inf
M_ST40SA	M_ST40SA	M_ST40SA	Inf
M_ST35TA	M_ST35TA	M_ST35TA	7.40
M_ST15CV	M_ST15CV	M_ST15CV	8.08
RD_ST129TA	RD_ST129TA	RD_ST129TA	5.55

prreg <- predictionStats_regression(cbind(TADPOLECrossMRITest$ADAS13,predict(bml,TADPOLECrossMRITest)),"ADAS13")

ADAS13

pander::pander(prreg)

• corci:

  cor
  0.725	0.688	0.758

• biasci: 1.010, 0.551 and 1.469

• RMSEci: 6.28, 5.97 and 6.63

• spearmanci:

  50%	2.5%	97.5%
  0.676	0.629	0.718

• MAEci:

  50%	2.5%	97.5%
  4.93	4.65	5.23

• pearson:

  Pearson’s product-moment correlation: predictions[, 1] and predictions[, 2]

  Test statistic	df	P value	Alternative hypothesis	cor
  27.9	702	8.24e-116 * * *	two.sided	0.725

par(op)

2.1.2 The formula network

cmax <- apply(bml$bagging$formulaNetwork,2,max)
cnames <- names(cmax[cmax>=0.250])
cmax <- cmax[cnames]

adma <- bml$bagging$formulaNetwork[cnames,cnames]

adma[adma<0.15] <- 0;
gr <- graph_from_adjacency_matrix(adma,mode = "undirected",diag = FALSE,weighted=TRUE)
gr$layout <- layout_with_fr

fc <- cluster_optimal(gr)
plot(fc, gr,
     edge.width=5*E(gr)$weight,
     vertex.size=20*cmax,
     vertex.label.cex=0.5,
     vertex.label.dist=0,
     main="ADAS13 Feature Association")

par(op)

2.1.3 The ADAS13 table

clusterFeatures <- fc$names

tableADAS13 <- sm$coefficients[clusterFeatures,
                                   c("Estimate",
                                     "lower",
                                     "mean",
                                     "upper",
                                     "model.MSE",
                                     "%VarExp",
                                     "F.pvalue",
                                     "Frequency")]

nugget <- fc$membership
names(nugget) <- clusterFeatures

tableADAS13$Cluster <- nugget[rownames(tableADAS13)]

rnames <- clusterFeatures[str_detect(clusterFeatures,"ST")]
frnames <- rnames
rnames <- str_replace_all(rnames,"M_","")
rnames <- str_replace_all(rnames,"RD_","")
description <- character()

for (ddet in c(1:length(rnames)))
{
  description <- c(description,TADPOLE_D1_D2_Dict$TEXT[str_detect(TADPOLE_D1_D2_Dict$FLDNAME,rnames[ddet])][1])
}
names(description) <- frnames

tableADAS13$Description <- description[rownames(tableADAS13)]
pander::pander(tableADAS13)

	Estimate	lower	mean	upper	model.MSE	%VarExp	F.pvalue	Frequency	Cluster	Description
Ventricles	-6.9694	-14.695	-6.9694	0.7568	46.5	0.00743	5.03e-04	0.45	1	NA
M_ST24TA	-264.3159	-421.053	-264.3159	-107.5783	46.9	0.02240	9.06e-09	0.95	1	Cortical Thickness Average of LeftEntorhinal
M_ST40CV	-75.5757	-140.070	-75.5757	-11.0816	46.9	0.01116	3.34e-05	1.00	1	Volume (Cortical Parcellation) of LeftMiddleTemporal
M_ST30SV	52.4153	15.133	52.4153	89.6976	46.9	0.01547	1.24e-06	1.00	1	Volume (WM Parcellation) of LeftInferiorLateralVentricle
Hippocampus	-149.7432	-210.157	-149.7432	-89.3298	47.8	0.03056	0.00e+00	1.00	1	NA
RD_ST31TA	13.3925	2.637	13.3925	24.1479	46.9	0.01250	1.12e-05	0.95	1	Cortical Thickness Average of LeftInferiorParietal
M_ST24CV	-283.8614	-369.063	-283.8614	-198.6597	48.9	0.06017	0.00e+00	1.00	2	Volume (Cortical Parcellation) of LeftEntorhinal
RD_ST40CV	12.6473	0.692	12.6473	24.6030	46.8	0.00954	1.03e-04	0.90	1	Volume (Cortical Parcellation) of LeftMiddleTemporal
RD_ST59SA	-12.2523	-23.944	-12.2523	-0.5609	47.9	0.00972	1.07e-04	0.80	1	Surface Area of LeftSupramarginal
M_ST60CV	85.1582	27.387	85.1582	142.9291	46.8	0.01670	5.42e-07	0.75	1	Volume (Cortical Parcellation) of LeftTemporalPole
AGE	-0.0676	-0.123	-0.0676	-0.0123	46.8	0.01195	1.68e-05	0.75	1	NA
M_ST129TS	734.2100	63.601	734.2100	1404.8191	48.5	0.01113	4.22e-05	0.75	2	Cortical Thickness Standard Deviation of LeftInsula
RD_ST56TA	12.0381	-7.646	12.0381	31.7221	50.0	0.00456	6.50e-03	0.40	2	Cortical Thickness Average of LeftSuperiorFrontal
M_ST62TA	176.8440	10.859	176.8440	342.8290	49.7	0.01064	7.82e-05	0.30	2	Cortical Thickness Average of LeftTransverseTemporal
M_ST31CV	-31.1982	-56.297	-31.1982	-6.0993	49.3	0.01230	1.96e-05	0.30	2	Volume (Cortical Parcellation) of LeftInferiorParietal
M_ST24SA	31.6447	6.909	31.6447	56.3805	49.8	0.01641	1.39e-06	0.25	2	Surface Area of LeftEntorhinal
M_ST60SA	38.0636	4.099	38.0636	72.0279	49.9	0.00971	1.45e-04	0.45	2	Surface Area of LeftTemporalPole
M_ST40TA	-350.0541	-597.735	-350.0541	-102.3734	50.6	0.01684	1.14e-06	0.25	2	Cortical Thickness Average of LeftMiddleTemporal
RD_ST57TA	14.1947	-1.982	14.1947	30.3715	49.9	0.00766	6.32e-04	0.40	2	Cortical Thickness Average of LeftSuperiorParietal

2.1.4 Cluster Predictions

cluster1names <- names(nugget)[nugget==1]

model1 <- lm(ADAS13~.,TADPOLECrossMRITrain[,c("ADAS13",cluster1names)])
#model1$coefficients <- bml$bagging$bagged.model$coefficients[c("(Intercept)",cluster1names)]

preo <- predict(bml,TADPOLECrossMRITest)
pre1 <- predict(model1,TADPOLECrossMRITest)
plot(preo,pre1)

pander::pander(cor.test(preo,TADPOLECrossMRITest$ADAS13,method="spearman"))

Spearman’s rank correlation rho: preo and TADPOLECrossMRITest$ADAS13

Test statistic	P value	Alternative hypothesis	rho
18789218	1.69e-95 * * *	two.sided	0.677

pander::pander(cor.test(pre1,TADPOLECrossMRITest$ADAS13,method="spearman"))

Spearman’s rank correlation rho: pre1 and TADPOLECrossMRITest$ADAS13

Test statistic	P value	Alternative hypothesis	rho
20653310	5.33e-84 * * *	two.sided	0.645

#cluster2names <- names(nugget)[nugget==2]

#model2 <- lm(ADAS13~.,TADPOLECrossMRITrain[,c("ADAS13",cluster2names)])
#model2$coefficients <- bml$bagging$bagged.model$coefficients[c("(Intercept)",cluster2names)]

#pre2 <- predict(model2,TADPOLECrossMRITest)
#plot(preo,pre2)
#plot(preo,0.5*(pre2+pre1))
#pander::pander(cor.test(pre2,TADPOLECrossMRITest$ADAS13,method="spearman"))

2.1.5 Decorrelating training and testing sets

TADPOLECrossMRITrainD <- GDSTMDecorrelation(TADPOLECrossMRITrain,Outcome="ADAS13",
                                        thr=0.6,
                                        type="RLM",
                                        method="spearman",
                                        verbose = TRUE)
#> 
#>  Included: 196 , Uni p: 0.01152849 To Outcome: 123 , Base: 6 , In Included: 6 , Base Cor: 15 
#> 1 , Top: 40 < 0.6 >( 2 )[ 1 : 0 : 0.594 ]( 39 , 92 , 0 ),<>Tot Used: 131 , Added: 92 , Zero Std: 0 , Max Cor: 0.9508271 
#> 2 , Top: 23 < 0.6 >( 1 )[ 1 : 0 : 0.594 ]( 23 , 36 , 39 ),<>Tot Used: 151 , Added: 36 , Zero Std: 0 , Max Cor: 0.9347139 
#> 3 , Top: 22 < 0.6 >( 2 )[ 1 : 0 : 0 ]( 22 , 26 , 60 ),<>Tot Used: 156 , Added: 26 , Zero Std: 0 , Max Cor: 0.8493966 
#> 4 , Top: 5 < 0.6 >( 3 )[ 1 : 0 : 0 ]( 5 , 7 , 78 ),<>Tot Used: 158 , Added: 7 , Zero Std: 0 , Max Cor: 0.5998317 
#> [ 5 ], 0.5967145 . Cor to Base: 101 , ABase: 66
TADPOLECrossMRITestD <-  predictDecorrelate(TADPOLECrossMRITrainD,TADPOLECrossMRITest)

2.2 Decorrelated

bmlD <- BSWiMS.model(ADAS13~.,TADPOLECrossMRITrainD,maxTrainModelSize=50,NumberofRepeats = 20)

[++-++–++-++-++-+–++-++-++-++-++-+–+–++-++-++-++-++-++-+++-]…

pander::pander(bmlD$bagging$Jaccard.SM)

0.314

fs <- bmlD$bagging$frequencyTable
barplot(fs[order(-fs)],las=2,main="Selected Features",cex.names = 0.5)

sm <- summary(bmlD)
pander::pander(sm$coefficients)

	Estimate	lower	mean	upper	u.MSE	r.MSE	model.MSE	NeRI	F.pvalue	t.pvalue	Sign.pvalue	Wilcox.pvalue	%VarExp	Frequency
Ba_Hippocampus	-227.2903	-2.59e+02	-227.2903	-1.95e+02	66.5	62.9	47.7	0.18221	0.00e+00	1.75e-19	6.62e-07	6.64e-10	0.15580	1.00
De_M_ST24TA	-350.0402	-4.50e+02	-350.0402	-2.51e+02	90.6	52.0	47.7	0.09009	0.00e+00	4.77e-06	8.25e-03	1.34e-03	0.04415	1.00
Ba_M_ST59TA	-947.5145	-1.08e+03	-947.5145	-8.14e+02	84.7	59.2	50.9	0.15279	0.00e+00	1.45e-11	2.43e-05	2.26e-06	0.08571	0.85
De_M_ST58CV	-477.0766	-5.58e+02	-477.0766	-3.96e+02	89.0	56.9	50.9	0.13838	0.00e+00	5.93e-12	1.15e-04	5.91e-07	0.06162	0.85
De_M_ST24CV	-184.2263	-2.66e+02	-184.2263	-1.02e+02	91.8	54.7	50.9	0.07466	6.45e-13	9.04e-06	2.35e-02	4.06e-03	0.03888	0.85
De_M_ST58TA	665.4390	2.67e+02	665.4390	1.06e+03	94.9	53.1	50.9	0.04162	2.26e-08	4.66e-03	1.21e-01	3.83e-02	0.02290	0.85
De_M_ST40SA	-33.5925	-5.55e+01	-33.5925	-1.17e+01	87.1	52.4	50.5	0.09426	2.42e-07	4.61e-03	5.92e-03	1.37e-02	0.01915	0.85
RD_ST31TA	16.2097	5.14e+00	16.2097	2.73e+01	89.3	49.4	47.7	0.08098	5.16e-07	2.56e-03	1.61e-02	2.26e-02	0.01698	1.00
M_ST60TS	568.3201	1.72e+02	568.3201	9.65e+02	93.0	52.7	51.0	0.05989	8.48e-07	6.31e-05	5.67e-02	9.53e-03	0.01747	0.65
De_M_ST45CV	591.1545	1.64e+02	591.1545	1.02e+03	94.3	52.5	50.9	0.11710	1.34e-06	1.74e-04	9.47e-04	2.71e-03	0.01672	0.85
De_M_ST30SV	53.6295	1.55e+01	53.6295	9.18e+01	81.0	49.2	47.7	0.07791	1.58e-06	8.41e-04	2.06e-02	1.15e-02	0.01540	1.00
RD_ST40TA	15.2471	4.18e+00	15.2471	2.63e+01	87.0	52.5	51.0	0.08461	1.94e-06	9.21e-03	1.28e-02	3.07e-02	0.01617	0.80
M_ST129TS	556.9061	1.15e+02	556.9061	9.98e+02	95.1	49.1	47.8	0.02944	9.54e-06	3.36e-02	2.01e-01	9.07e-02	0.01309	0.90
De_M_ST51CV	27.5007	4.87e+00	27.5007	5.01e+01	96.5	51.7	50.5	0.03837	1.59e-05	3.00e-02	1.57e-01	9.60e-02	0.01292	0.20
De_M_ST31TA	-634.3612	-1.17e+03	-634.3612	-1.02e+02	93.0	48.9	47.7	0.03571	2.10e-05	2.04e-02	1.65e-01	1.11e-01	0.01187	1.00
De_M_ST59TS	241.1974	3.51e+01	241.1974	4.47e+02	94.9	52.5	51.3	0.00142	3.06e-05	3.26e-02	5.00e-01	1.54e-01	0.01222	0.10
Ba_RD_ST59SA	-11.5703	-2.19e+01	-11.5703	-1.21e+00	97.0	48.8	47.7	0.01684	5.55e-05	1.65e-01	3.33e-01	3.16e-01	0.01055	1.00
Ba_ICV	-0.0283	-5.38e-02	-0.0283	-2.73e-03	96.7	51.6	50.5	0.05769	6.26e-05	1.42e-03	6.66e-02	2.40e-02	0.01099	0.20
Ba_M_ST13SA	-3.7356	-7.13e+00	-3.7356	-3.44e-01	91.2	53.0	51.9	0.05204	7.83e-05	1.03e-01	8.31e-02	1.14e-01	0.01113	0.15
Ba_RD_ST40CV	13.3591	9.90e-01	13.3591	2.57e+01	90.6	48.8	47.8	0.05825	8.21e-05	1.11e-02	6.02e-02	5.50e-02	0.01004	1.00
De_M_ST60CV	29.0941	1.24e+00	29.0941	5.69e+01	88.6	49.4	48.5	0.04662	9.99e-05	3.50e-02	9.46e-02	1.58e-01	0.00990	0.25
De_M_ST52TA	-248.6152	-4.89e+02	-248.6152	-8.06e+00	95.7	51.3	50.4	0.04247	1.34e-04	7.81e-03	1.17e-01	3.01e-02	0.00989	0.40
De_M_ST55SA	-2.1791	-4.29e+00	-2.1791	-6.43e-02	94.5	48.6	47.7	0.07539	1.48e-04	5.83e-02	2.49e-02	1.76e-01	0.00922	0.10
Ba_M_ST32SA	-1.0560	-1.68e+00	-1.0560	-4.35e-01	88.8	53.5	52.3	0.03694	1.50e-04	3.30e-03	1.45e-01	5.13e-02	0.01210	0.10
AGE	-0.0307	-6.10e-02	-0.0307	-4.13e-04	95.5	48.7	47.8	0.04952	1.77e-04	3.48e-03	9.71e-02	6.47e-02	0.00902	0.45
De_M_ST31SA	-4.9088	-9.76e+00	-4.9088	-5.61e-02	91.0	48.3	47.5	0.01733	1.81e-04	2.25e-01	3.49e-01	2.82e-01	0.00892	0.35
RD_ST57TA	13.5792	-1.58e-01	13.5792	2.73e+01	91.7	51.4	50.4	0.07524	2.06e-04	2.40e-02	2.40e-02	2.64e-02	0.00935	0.75
De_M_ST56CV	-16.9969	-3.42e+01	-16.9969	1.76e-01	95.0	50.5	49.7	0.00385	2.55e-04	1.16e-01	5.05e-01	2.30e-01	0.00885	0.15
RD_ST14TS	-0.3544	-7.13e-01	-0.3544	3.87e-03	95.6	52.2	51.3	-0.03003	2.59e-04	1.24e-01	1.00e+00	4.14e-01	0.00915	0.15
De_M_ST60SA	52.4690	-4.66e+00	52.4690	1.10e+02	96.0	51.5	50.7	0.02702	4.72e-04	1.66e-02	2.27e-01	1.24e-01	0.00819	0.60
RD_ST56TA	13.2876	-1.37e+00	13.2876	2.79e+01	94.4	49.1	48.3	-0.01024	5.61e-04	2.13e-01	6.82e-01	3.79e-01	0.00757	0.75
De_M_ST32CV	-5.8744	-1.11e+01	-5.8744	-6.13e-01	93.2	54.4	53.6	0.04761	1.48e-03	3.95e-02	1.03e-01	1.14e-01	0.00763	0.10
De_M_ST62TA	80.9840	-1.62e+01	80.9840	1.78e+02	95.6	50.6	50.1	0.01218	3.10e-03	9.61e-02	3.79e-01	1.59e-01	0.00557	0.35
De_RD_ST31CV	2.3377	-9.76e-01	2.3377	5.65e+00	95.9	51.6	51.0	0.03105	3.29e-03	2.03e-01	2.07e-01	2.09e-01	0.00554	0.10
M_ST24TS	195.1379	-6.28e+01	195.1379	4.53e+02	92.4	51.2	50.7	0.01437	4.56e-03	1.16e-01	3.66e-01	1.72e-01	0.00515	0.50
De_M_ST40CV	39.5830	-2.21e+01	39.5830	1.01e+02	93.3	51.0	50.5	0.01157	5.53e-03	2.32e-01	4.11e-01	2.54e-01	0.00481	0.25
Ba_Ventricles	0.7751	-5.55e-01	0.7751	2.11e+00	86.3	47.8	47.4	0.07397	8.60e-03	4.97e-02	2.71e-02	7.81e-02	0.00395	0.10
De_M_ST55CV	-27.7909	-6.20e+01	-27.7909	6.37e+00	90.3	51.3	50.9	0.01664	8.92e-03	2.49e-01	3.27e-01	2.91e-01	0.00442	0.40
De_M_ST26CV	-12.2219	-3.57e+01	-12.2219	1.12e+01	94.8	52.5	52.1	0.02459	1.62e-02	9.30e-02	2.60e-01	1.56e-01	0.00351	0.15
RD_ST24TA	0.3763	-3.84e-01	0.3763	1.14e+00	89.5	51.3	51.0	0.05436	1.88e-02	1.57e-01	7.86e-02	1.22e-01	0.00323	0.10

pander::pander(bmlD$univariate[bmlD$selectedfeatures,])

	Name	RName	ZUni
Ba_Hippocampus	Ba_Hippocampus	Ba_Hippocampus	Inf
Ba_Ventricles	Ba_Ventricles	Ba_Ventricles	Inf
De_M_ST30SV	De_M_ST30SV	De_M_ST30SV	Inf
De_M_ST24TA	De_M_ST24TA	De_M_ST24TA	7.02
Ba_RD_ST59SA	Ba_RD_ST59SA	Ba_RD_ST59SA	1.15
De_M_ST31TA	De_M_ST31TA	De_M_ST31TA	5.52
RD_ST31TA	RD_ST31TA	RD_ST31TA	7.71
M_ST129TS	M_ST129TS	M_ST129TS	3.87
De_M_ST51TA	De_M_ST51TA	De_M_ST51TA	2.33
De_M_ST55SA	De_M_ST55SA	De_M_ST55SA	4.38
AGE	AGE	AGE	3.47
De_M_ST60CV	De_M_ST60CV	De_M_ST60CV	8.04
Ba_RD_ST40CV	Ba_RD_ST40CV	Ba_RD_ST40CV	7.01
De_M_ST31SA	De_M_ST31SA	De_M_ST31SA	6.76
RD_ST56TA	RD_ST56TA	RD_ST56TA	4.51
De_M_ST24CV	De_M_ST24CV	De_M_ST24CV	6.31
De_M_ST40SA	De_M_ST40SA	De_M_ST40SA	Inf
Ba_M_ST59TA	Ba_M_ST59TA	Ba_M_ST59TA	Inf
De_M_ST40CV	De_M_ST40CV	De_M_ST40CV	5.35
De_M_ST62TA	De_M_ST62TA	De_M_ST62TA	3.32
De_M_ST55CV	De_M_ST55CV	De_M_ST55CV	7.15
RD_ST24TA	RD_ST24TA	RD_ST24TA	7.57
De_M_ST58CV	De_M_ST58CV	De_M_ST58CV	7.84
RD_ST57TA	RD_ST57TA	RD_ST57TA	6.36
De_M_ST58TA	De_M_ST58TA	De_M_ST58TA	4.07
Ba_M_ST24SA	Ba_M_ST24SA	Ba_M_ST24SA	3.74
De_M_ST60SA	De_M_ST60SA	De_M_ST60SA	2.89
M_ST24TS	M_ST24TS	M_ST24TS	5.93
RD_ST40TA	RD_ST40TA	RD_ST40TA	Inf
Ba_M_ST32SA	Ba_M_ST32SA	Ba_M_ST32SA	7.93
De_M_ST26CV	De_M_ST26CV	De_M_ST26CV	4.17
De_M_ST51CV	De_M_ST51CV	De_M_ST51CV	2.11
M_ST60TS	M_ST60TS	M_ST60TS	5.52
De_RD_ST31CV	De_RD_ST31CV	De_RD_ST31CV	3.04
De_M_ST32CV	De_M_ST32CV	De_M_ST32CV	5.44
Ba_M_ST59SA	Ba_M_ST59SA	Ba_M_ST59SA	4.24
De_M_ST45CV	De_M_ST45CV	De_M_ST45CV	4.54
Ba_M_ST13SA	Ba_M_ST13SA	Ba_M_ST13SA	6.66
De_M_ST59TS	De_M_ST59TS	De_M_ST59TS	4.06
Ba_ICV	Ba_ICV	Ba_ICV	1.86
De_M_ST52TA	De_M_ST52TA	De_M_ST52TA	3.25
De_M_ST56CV	De_M_ST56CV	De_M_ST56CV	4.00
Ba_M_ST15SA	Ba_M_ST15SA	Ba_M_ST15SA	3.08
RD_ST14TS	RD_ST14TS	RD_ST14TS	3.38
RD_ST52TA	RD_ST52TA	RD_ST52TA	5.41
M_ST54TS	M_ST54TS	M_ST54TS	3.75
RD_ST32TA	RD_ST32TA	RD_ST32TA	7.18
De_M_ST11SV	De_M_ST11SV	De_M_ST11SV	Inf
M_ST44TA	M_ST44TA	M_ST44TA	Inf
De_M_ST44CV	De_M_ST44CV	De_M_ST44CV	1.82
M_ST43TS	M_ST43TS	M_ST43TS	2.82
M_ST44SA	M_ST44SA	M_ST44SA	4.91
De_ST6SV	De_ST6SV	De_ST6SV	1.47

prreg <- predictionStats_regression(cbind(TADPOLECrossMRITestD$ADAS13,predict(bmlD,TADPOLECrossMRITestD)),"ADAS13")

ADAS13

pander::pander(prreg)

• corci:

  cor
  0.715	0.677	0.75

• biasci: 1.064, 0.599 and 1.529

• RMSEci: 6.38, 6.06 and 6.73

• spearmanci:

  50%	2.5%	97.5%
  0.677	0.63	0.718

• MAEci:

  50%	2.5%	97.5%
  4.98	4.69	5.29

• pearson:

  Pearson’s product-moment correlation: predictions[, 1] and predictions[, 2]

  Test statistic	df	P value	Alternative hypothesis	cor
  27.1	702	2.08e-111 * * *	two.sided	0.715

par(op)

2.2.1 The formula network

cmax <- apply(bmlD$bagging$formulaNetwork,2,max)
cnames <- names(cmax[cmax>=0.250])
cmax <- cmax[cnames]

adma <- bmlD$bagging$formulaNetwork[cnames,cnames]

adma[adma<0.15] <- 0;
gr <- graph_from_adjacency_matrix(adma,mode = "undirected",diag = FALSE,weighted=TRUE)
gr$layout <- layout_with_fr

fc <- cluster_optimal(gr)
plot(fc, gr,
     edge.width=5*E(gr)$weight,
     vertex.size=20*cmax,
     vertex.label.cex=0.5,
     vertex.label.dist=0,
     main="ADAS13 Feature Association")

par(op)

2.2.2 Decorrelated ADAS13 table

clusterFeatures <- fc$names

tableADAS13D <- sm$coefficients[clusterFeatures,
                                   c("Estimate",
                                     "lower",
                                     "mean",
                                     "upper",
                                     "model.MSE",
                                     "%VarExp",
                                     "F.pvalue",
                                     "Frequency")]

nugget <- fc$membership
names(nugget) <- clusterFeatures

tableADAS13D$Cluster <- nugget[rownames(tableADAS13D)]

rnames <- clusterFeatures[str_detect(clusterFeatures,"ST")]
frnames <- rnames
rnames <- str_replace_all(rnames,"M_","")
rnames <- str_replace_all(rnames,"RD_","")
rnames <- str_replace_all(rnames,"Ba_","")
rnames <- str_replace_all(rnames,"De_","")
description <- character()

for (ddet in c(1:length(rnames)))
{
  description <- c(description,TADPOLE_D1_D2_Dict$TEXT[str_detect(TADPOLE_D1_D2_Dict$FLDNAME,rnames[ddet])][1])
}
names(description) <- frnames

tableADAS13D$Description <- description[rownames(tableADAS13D)]

## Getting the decorrelation formula
dc <- getDerivedCoefficients(TADPOLECrossMRITrainD)
decornames <- rownames(sm$coefficients)

deNames_in_dc <- decornames[decornames %in% names(dc)]
theDeFormulas <- dc[deNames_in_dc]
deFromula <- character(length(theDeFormulas))
names(deFromula) <- names(theDeFormulas)
for (dx in names(deFromula))
{
  coef <- theDeFormulas[[dx]]
  cname <- names(theDeFormulas[[dx]])
  names(cname) <- cname
  for (cf in names(coef))
  {
    if (cf != dx)
    {
      if (coef[cf]>0)
      {
        deFromula[dx] <- paste(deFromula[dx],
                               sprintf("+ %5.3f*%s",coef[cf],cname[cf]))
      }
      else
      {
        deFromula[dx] <- paste(deFromula[dx],
                               sprintf("%5.3f*%s",coef[cf],cname[cf]))
      }
    }
  }
}

tableADAS13D$DecorFormula <- deFromula[rownames(tableADAS13D)]


pander::pander(tableADAS13D)

	Estimate	lower	mean	upper	model.MSE	%VarExp	F.pvalue	Frequency	Cluster	Description	DecorFormula
Ba_Hippocampus	-227.2903	-259.366	-227.2903	-1.95e+02	47.7	0.15580	0.00e+00	1.00	1	NA	NA
De_M_ST30SV	53.6295	15.490	53.6295	9.18e+01	47.7	0.01540	1.58e-06	1.00	1	Volume (WM Parcellation) of LeftInferiorLateralVentricle	-0.320Ventricles + 1.000M_ST30SV
De_M_ST24TA	-350.0402	-449.510	-350.0402	-2.51e+02	47.7	0.04415	0.00e+00	1.00	1	Cortical Thickness Average of LeftEntorhinal	-0.313Hippocampus + 1.000M_ST24TA
Ba_RD_ST59SA	-11.5703	-21.930	-11.5703	-1.21e+00	47.7	0.01055	5.55e-05	1.00	1	Surface Area of LeftSupramarginal	NA
De_M_ST31TA	-634.3612	-1167.057	-634.3612	-1.02e+02	47.7	0.01187	2.10e-05	1.00	1	Cortical Thickness Average of LeftInferiorParietal	+ 1.000M_ST31TA -0.917M_ST59TA
RD_ST31TA	16.2097	5.138	16.2097	2.73e+01	47.7	0.01698	5.16e-07	1.00	1	Cortical Thickness Average of LeftInferiorParietal	NA
M_ST129TS	556.9061	115.397	556.9061	9.98e+02	47.8	0.01309	9.54e-06	0.90	1	Cortical Thickness Standard Deviation of LeftInsula	NA
AGE	-0.0307	-0.061	-0.0307	-4.13e-04	47.8	0.00902	1.77e-04	0.45	1	NA	NA
De_M_ST60CV	29.0941	1.245	29.0941	5.69e+01	48.5	0.00990	9.99e-05	0.25	1	Volume (Cortical Parcellation) of LeftTemporalPole	NA
Ba_RD_ST40CV	13.3591	0.990	13.3591	2.57e+01	47.8	0.01004	8.21e-05	1.00	1	Volume (Cortical Parcellation) of LeftMiddleTemporal	NA
De_M_ST31SA	-4.9088	-9.762	-4.9088	-5.61e-02	47.5	0.00892	1.81e-04	0.35	1	Surface Area of LeftInferiorParietal	NA
RD_ST56TA	13.2876	-1.370	13.2876	2.79e+01	48.3	0.00757	5.61e-04	0.75	1	Cortical Thickness Average of LeftSuperiorFrontal	NA
De_M_ST24CV	-184.2263	-266.118	-184.2263	-1.02e+02	50.9	0.03888	6.45e-13	0.85	2	Volume (Cortical Parcellation) of LeftEntorhinal	-0.520Hippocampus + 1.000M_ST24CV
De_M_ST40SA	-33.5925	-55.484	-33.5925	-1.17e+01	50.5	0.01915	2.42e-07	0.85	2	Surface Area of LeftMiddleTemporal	NA
Ba_M_ST59TA	-947.5145	-1081.229	-947.5145	-8.14e+02	50.9	0.08571	0.00e+00	0.85	2	Cortical Thickness Average of LeftSupramarginal	NA
De_M_ST40CV	39.5830	-22.137	39.5830	1.01e+02	50.5	0.00481	5.53e-03	0.25	2	Volume (Cortical Parcellation) of LeftMiddleTemporal	-0.125Hippocampus -3.505M_ST40TA + 0.627M_ST59TA -0.204M_ST40SA + 1.000*M_ST40CV
De_M_ST62TA	80.9840	-16.225	80.9840	1.78e+02	50.1	0.00557	3.10e-03	0.35	1	Cortical Thickness Average of LeftTransverseTemporal	-0.910M_ST59TA + 1.000M_ST62TA
De_M_ST55CV	-27.7909	-61.952	-27.7909	6.37e+00	50.9	0.00442	8.92e-03	0.40	2	Volume (Cortical Parcellation) of LeftRostralMiddleFrontal	-0.151M_ST55SA -0.911M_ST25CV + 1.000*M_ST55CV
De_M_ST58CV	-477.0766	-558.238	-477.0766	-3.96e+02	50.9	0.06162	0.00e+00	0.85	2	Volume (Cortical Parcellation) of LeftSuperiorTemporal	+ 0.243Hippocampus -1.959M_ST58TA -2.029M_ST59TA -0.239M_ST58SA + 1.000*M_ST58CV
RD_ST57TA	13.5792	-0.158	13.5792	2.73e+01	50.4	0.00935	2.06e-04	0.75	2	Cortical Thickness Average of LeftSuperiorParietal	NA
De_M_ST58TA	665.4390	266.958	665.4390	1.06e+03	50.9	0.02290	2.26e-08	0.85	2	Cortical Thickness Average of LeftSuperiorTemporal	-0.124Hippocampus + 1.000M_ST58TA -0.545*M_ST59TA
De_M_ST60SA	52.4690	-4.663	52.4690	1.10e+02	50.7	0.00819	4.72e-04	0.60	2	Surface Area of LeftTemporalPole	-0.472Hippocampus + 2.343M_ST60TA + 1.000M_ST60SA -1.296M_ST60CV
M_ST24TS	195.1379	-62.843	195.1379	4.53e+02	50.7	0.00515	4.56e-03	0.50	2	Cortical Thickness Standard Deviation of LeftEntorhinal	NA
RD_ST40TA	15.2471	4.179	15.2471	2.63e+01	51.0	0.01617	1.94e-06	0.80	2	Cortical Thickness Average of LeftMiddleTemporal	NA
M_ST60TS	568.3201	171.561	568.3201	9.65e+02	51.0	0.01747	8.48e-07	0.65	2	Cortical Thickness Standard Deviation of LeftTemporalPole	NA
De_M_ST45CV	591.1545	164.151	591.1545	1.02e+03	50.9	0.01672	1.34e-06	0.85	2	Volume (Cortical Parcellation) of LeftParsOpercularis	-2.430M_ST45TA -0.075M_ST59TA -0.259M_ST45SA + 1.000M_ST45CV
De_M_ST52TA	-248.6152	-489.171	-248.6152	-8.06e+00	50.4	0.00989	1.34e-04	0.40	2	Cortical Thickness Average of LeftPrecuneus	+ 1.000M_ST52TA -0.847M_ST59TA

3 Diagnosis MCI vs AD

3.0.1 the set

TrainFraction <- 0.60;

TADPOLECrossMRI <- subset(validBaselineTadpole,DX == "Dementia" | DX == "MCI")
table(TADPOLECrossMRI$DX)

Dementia MCI NL 244 711 0

TADPOLECrossMRI$DX <- 1*(as.character(TADPOLECrossMRI$DX) == "Dementia")
table(TADPOLECrossMRI$DX)

0 1 711 244

TADPOLECrossMRI$ADAS13 <- NULL
TADPOLECrossMRI$ADAS11 <- NULL
TADPOLECrossMRI$MMSE <- NULL
TADPOLECrossMRI$RAVLT_immediate <- NULL
TADPOLECrossMRI$RAVLT_learning <- NULL
TADPOLECrossMRI$RAVLT_perc_forgetting <- NULL
TADPOLECrossMRI$RAVLT_forgetting <- NULL
TADPOLECrossMRI$FAQ <- NULL

TADPOLE_Cases <- subset(TADPOLECrossMRI,DX==1)
TADPOLE_Controls <- subset(TADPOLECrossMRI,DX==0)
trainCasesSet <- sample(nrow(TADPOLE_Cases),nrow(TADPOLE_Cases)*TrainFraction)
trainControlSet <- sample(nrow(TADPOLE_Controls),nrow(TADPOLE_Controls)*TrainFraction)

TADPOLE_DX_TRAIN <- rbind(TADPOLE_Cases[trainCasesSet,],TADPOLE_Controls[trainControlSet,])
TADPOLE_DX_TEST <- TADPOLECrossMRI[!(rownames(TADPOLECrossMRI) %in% rownames(TADPOLE_DX_TRAIN)),]
table(TADPOLE_DX_TEST$DX)

0 1 285 98

par(op)

3.0.2 Learning

bDXml <- BSWiMS.model(DX~.,TADPOLE_DX_TRAIN,NumberofRepeats = 20)

[++++++-+++-+++++-++++++–++++-+++++-++++++-++++–++++-++-++++++-++++-++-+++++–+++++-+-++++-+-+++++++++-++++++++-++-+++++-++++++++–++++-+-++++-]………..

pander::pander(bDXml$bagging$Jaccard.SM)

0.0902

fs <- bDXml$bagging$frequencyTable
barplot(fs[order(-fs)],las=2,main="Selected Features",cex.names = 0.5)

sm <- summary(bDXml)
pander::pander(sm$coefficients)

	Estimate	lower	OR	upper	u.Accuracy	r.Accuracy	full.Accuracy	u.AUC	r.AUC	full.AUC	IDI	NRI	z.IDI	z.NRI	Delta.AUC	Frequency
M_ST40TA	-79.016	1.07e-40	4.83e-35	2.18e-29	0.661	0.630	0.717	0.655	0.631	0.718	0.1247	0.689	11.13	10.78	8.71e-02	0.65
M_ST32TA	-30.317	3.08e-16	6.81e-14	1.51e-11	0.639	0.649	0.707	0.645	0.638	0.705	0.1144	0.622	10.34	9.62	6.70e-02	0.40
M_ST31TA	-70.265	1.23e-36	3.05e-31	7.54e-26	0.636	0.678	0.728	0.626	0.681	0.731	0.0896	0.574	9.09	8.80	5.00e-02	0.50
M_ST24CV	-20.188	3.20e-11	1.71e-09	9.12e-08	0.678	0.689	0.723	0.660	0.684	0.724	0.0807	0.554	8.64	8.51	3.97e-02	0.95
Hippocampus	-13.711	5.93e-08	1.11e-06	2.08e-05	0.667	0.690	0.723	0.672	0.684	0.730	0.0787	0.549	8.64	8.41	4.61e-02	1.00
M_ST52CV	-1.379	1.84e-01	2.52e-01	3.44e-01	0.616	0.646	0.679	0.602	0.635	0.685	0.0795	0.419	8.45	6.28	4.94e-02	0.10
M_ST32CV	-14.298	2.01e-08	6.17e-07	1.90e-05	0.694	0.704	0.735	0.693	0.699	0.739	0.0710	0.518	8.11	7.84	4.00e-02	1.00
M_ST30SV	2.881	8.13e+00	1.78e+01	3.91e+01	0.651	0.686	0.700	0.652	0.682	0.705	0.0709	0.487	7.72	7.40	2.35e-02	0.90
M_ST40SA	-0.832	3.54e-01	4.35e-01	5.35e-01	0.600	0.667	0.693	0.594	0.655	0.696	0.0626	0.469	7.54	7.06	4.14e-02	0.15
M_ST12SV	-10.998	5.36e-07	1.67e-05	5.23e-04	0.666	0.692	0.717	0.671	0.693	0.722	0.0599	0.551	7.36	8.41	2.87e-02	0.90
M_ST40CV	-10.650	9.62e-07	2.37e-05	5.84e-04	0.693	0.694	0.716	0.699	0.692	0.718	0.0570	0.439	7.10	6.60	2.55e-02	0.95
M_ST31CV	-9.623	3.00e-06	6.62e-05	1.46e-03	0.653	0.697	0.724	0.643	0.698	0.730	0.0596	0.451	7.02	6.80	3.22e-02	0.90
M_ST24TA	-24.833	6.09e-15	1.64e-11	4.42e-08	0.691	0.710	0.735	0.681	0.710	0.739	0.0547	0.522	7.02	7.92	2.86e-02	1.00
M_ST44CV	-3.787	6.08e-03	2.27e-02	8.44e-02	0.617	0.681	0.709	0.618	0.673	0.708	0.0489	0.396	6.52	5.92	3.56e-02	0.20
M_ST60TA	-5.852	4.15e-04	2.87e-03	1.99e-02	0.634	0.680	0.706	0.619	0.672	0.704	0.0471	0.456	6.49	6.92	3.23e-02	0.25
M_ST45TA	32.731	5.97e+07	1.64e+14	4.51e+20	0.582	0.682	0.712	0.570	0.678	0.708	0.0449	0.400	6.40	5.99	2.94e-02	0.40
M_ST55TS	115.895	1.13e+29	2.15e+50	4.08e+71	0.555	0.685	0.713	0.547	0.683	0.715	0.0439	0.457	6.19	6.89	3.17e-02	0.40
M_ST55SA	-0.886	2.76e-01	4.12e-01	6.16e-01	0.598	0.687	0.701	0.597	0.683	0.696	0.0408	0.417	5.86	6.23	1.37e-02	0.30
M_ST32SA	-2.990	9.67e-03	5.03e-02	2.61e-01	0.622	0.700	0.721	0.622	0.708	0.727	0.0380	0.408	5.64	6.10	1.96e-02	0.90
RD_ST31TA	0.846	1.49e+00	2.33e+00	3.65e+00	0.626	0.685	0.701	0.574	0.691	0.702	0.0369	0.285	5.54	4.31	1.13e-02	0.55
M_ST39CV	9.367	5.56e+01	1.17e+04	2.46e+06	0.563	0.711	0.729	0.562	0.716	0.734	0.0342	0.332	5.54	4.96	1.76e-02	0.65
M_ST24SA	2.887	3.23e+00	1.79e+01	9.97e+01	0.535	0.711	0.724	0.550	0.709	0.724	0.0323	0.283	5.51	4.19	1.54e-02	0.25
M_ST129TS	79.463	8.02e+17	3.24e+34	1.31e+51	0.605	0.704	0.714	0.580	0.701	0.715	0.0379	0.366	5.47	5.47	1.35e-02	0.70
M_ST46TS	107.131	2.05e+18	3.36e+46	5.51e+74	0.583	0.722	0.733	0.560	0.725	0.737	0.0320	0.322	5.36	4.78	1.25e-02	1.00
M_ST43TS	131.338	9.38e+24	1.09e+57	1.28e+89	0.575	0.702	0.717	0.553	0.702	0.720	0.0341	0.337	5.33	5.02	1.73e-02	0.90
M_ST60TS	65.901	2.55e+12	4.17e+28	6.83e+44	0.594	0.704	0.727	0.591	0.709	0.733	0.0308	0.358	5.31	5.33	2.36e-02	0.95
M_ST35SA	-3.534	3.29e-03	2.92e-02	2.59e-01	0.566	0.707	0.732	0.581	0.711	0.736	0.0293	0.396	5.26	5.92	2.53e-02	0.95
M_ST59TA	-2.806	1.02e-02	6.05e-02	3.59e-01	0.610	0.701	0.713	0.602	0.692	0.707	0.0329	0.196	5.24	2.88	1.52e-02	0.10
M_ST36TS	20.552	3.37e+03	8.43e+08	2.11e+14	0.559	0.697	0.715	0.531	0.687	0.707	0.0317	0.385	5.16	5.77	2.01e-02	0.15
M_ST62TA	3.333	2.35e+00	2.80e+01	3.34e+02	0.536	0.689	0.710	0.520	0.696	0.711	0.0276	0.280	5.07	4.13	1.47e-02	0.10
M_ST59CV	-2.419	2.11e-02	8.90e-02	3.75e-01	0.651	0.705	0.711	0.649	0.697	0.709	0.0322	0.393	5.06	5.87	1.11e-02	0.20
M_ST26CV	-4.445	1.04e-03	1.17e-02	1.32e-01	0.660	0.707	0.717	0.647	0.701	0.715	0.0307	0.336	5.02	5.01	1.40e-02	0.55
RD_ST44TA	0.503	1.18e+00	1.65e+00	2.32e+00	0.608	0.699	0.713	0.574	0.700	0.712	0.0281	0.301	4.93	4.51	1.28e-02	0.50
M_ST43TA	20.526	2.83e+02	8.21e+08	2.38e+15	0.552	0.696	0.715	0.541	0.694	0.715	0.0272	0.263	4.78	3.88	2.10e-02	0.25
M_ST14TA	3.715	3.48e+00	4.10e+01	4.84e+02	0.540	0.682	0.713	0.540	0.676	0.712	0.0270	0.292	4.76	4.32	3.66e-02	0.20
RD_ST52TA	1.696	1.76e+00	5.45e+00	1.69e+01	0.616	0.716	0.719	0.556	0.720	0.720	0.0281	0.220	4.74	3.31	-3.62e-05	0.75
M_ST54CV	1.300	1.47e+00	3.67e+00	9.14e+00	0.530	0.734	0.745	0.526	0.737	0.757	0.0219	0.242	4.63	3.57	2.04e-02	0.10
M_ST44TS	-53.998	1.21e-41	3.54e-24	1.03e-06	0.541	0.705	0.715	0.542	0.702	0.717	0.0234	0.306	4.47	4.53	1.44e-02	0.55

pander::pander(bDXml$univariate[bDXml$selectedfeatures,])

	Name	RName	ZUni
M_ST32CV	M_ST32CV	M_ST32CV	9.01
M_ST24TA	M_ST24TA	M_ST24TA	9.11
M_ST46TS	M_ST46TS	M_ST46TS	2.82
M_ST30SV	M_ST30SV	M_ST30SV	8.41
M_ST32SA	M_ST32SA	M_ST32SA	6.46
Hippocampus	Hippocampus	Hippocampus	8.49
M_ST12SV	M_ST12SV	M_ST12SV	8.10
M_ST31CV	M_ST31CV	M_ST31CV	7.16
M_ST54CV	M_ST54CV	M_ST54CV	1.78
M_ST60TS	M_ST60TS	M_ST60TS	4.29
RD_ST31TA	RD_ST31TA	RD_ST31TA	4.55
M_ST40CV	M_ST40CV	M_ST40CV	7.85
M_ST39CV	M_ST39CV	M_ST39CV	2.60
M_ST24CV	M_ST24CV	M_ST24CV	7.76
RD_ST52TA	RD_ST52TA	RD_ST52TA	4.65
M_ST35CV	M_ST35CV	M_ST35CV	6.11
M_ST43TS	M_ST43TS	M_ST43TS	2.94
M_ST35SA	M_ST35SA	M_ST35SA	3.29
M_ST129TS	M_ST129TS	M_ST129TS	3.90
M_ST24SA	M_ST24SA	M_ST24SA	1.81
M_ST31TA	M_ST31TA	M_ST31TA	6.73
M_ST26CV	M_ST26CV	M_ST26CV	7.18
M_ST40TA	M_ST40TA	M_ST40TA	7.00
RD_ST44TA	RD_ST44TA	RD_ST44TA	3.73
M_ST56TS	M_ST56TS	M_ST56TS	3.07
M_ST45TA	M_ST45TA	M_ST45TA	3.22
M_ST32TA	M_ST32TA	M_ST32TA	7.04
M_ST44CV	M_ST44CV	M_ST44CV	6.34
M_ST44TS	M_ST44TS	M_ST44TS	1.41
M_ST59TA	M_ST59TA	M_ST59TA	6.44
RD_ST29SV	RD_ST29SV	RD_ST29SV	3.65
M_ST43TA	M_ST43TA	M_ST43TA	2.56
M_ST55SA	M_ST55SA	M_ST55SA	4.37
RD_ST55TA	RD_ST55TA	RD_ST55TA	3.01
M_ST40SA	M_ST40SA	M_ST40SA	4.50
M_ST60TA	M_ST60TA	M_ST60TA	5.72
M_ST59CV	M_ST59CV	M_ST59CV	6.65
M_ST55TS	M_ST55TS	M_ST55TS	1.42
M_ST14TA	M_ST14TA	M_ST14TA	1.67
M_ST26TA	M_ST26TA	M_ST26TA	6.91
M_ST62TA	M_ST62TA	M_ST62TA	2.33
M_ST52CV	M_ST52CV	M_ST52CV	6.53
M_ST36TS	M_ST36TS	M_ST36TS	1.62
M_ST55CV	M_ST55CV	M_ST55CV	6.18
M_ST23TA	M_ST23TA	M_ST23TA	2.49
M_ST31SA	M_ST31SA	M_ST31SA	3.88
M_ST58TA	M_ST58TA	M_ST58TA	6.62

prBin <- predictionStats_binary(cbind(TADPOLE_DX_TEST$DX,predict(bDXml,TADPOLE_DX_TEST)),"MCI vs Dementia")

MCI vs Dementia

pander::pander(prBin$aucs)

est	lower	upper
0.821	0.775	0.867

pander::pander(prBin$accc)

est	lower	upper
0.734	0.686	0.777

pander::pander(prBin$berror)

50%	2.5%	97.5%
0.255	0.209	0.305

pander::pander(prBin$sensitivity)

est	lower	upper
0.765	0.669	0.845

par(op)

plot(sm$coefficients$IDI,sm$coefficients$Delta.AUC)

plot(sm$coefficients$NRI,sm$coefficients$Delta.AUC)

par(op)
gplots::heatmap.2(log(100*bDXml$bagging$formulaNetwork+1),
                  trace = "none",
                  scale = "none",
                  mar = c(10,10),
                  col=rev(heat.colors(5)),
                  main = "formulaNetwork",
                  cexRow = 0.5,
                  cexCol = 0.5,
                  key.title=NA,
                  key.xlab="Spearman Correlation",
                  xlab="Feature", ylab="Feature")
par(op)

gplots::heatmap.2(log(100*bDXml$bagging$WformulaNetwork+1),
                  trace = "none",
                  scale = "none",
                  mar = c(10,10),
                  col=rev(heat.colors(5)),
                  main = "Wnetwork",
                  cexRow = 0.5,
                  cexCol = 0.5,
                  key.title=NA,
                  key.xlab="Spearman Correlation",
                  xlab="Feature", ylab="Feature")

par(op)

3.0.3 The formula network

cmax <- apply(bDXml$bagging$formulaNetwork,2,max)
cnames <- names(cmax[cmax>=0.25])
cmax <- cmax[cnames]

adma <- bDXml$bagging$formulaNetwork[cnames,cnames]

adma[adma<0.15] <- 0;
gr <- graph_from_adjacency_matrix(adma,mode = "undirected",diag = FALSE,weighted=TRUE)
gr$layout <- layout_with_fr

fc <- cluster_optimal(gr)
plot(fc, gr,
     edge.width=10*E(gr)$weight,
     vertex.size=10*cmax,
     vertex.label.cex=0.75,
     vertex.label.dist=0,
     main="MCI vs Dementia Diagnosis")

par(op)

3.0.4 MCI vs AD table

clusterFeatures <- fc$names

tableMCI_De <- sm$coefficients[clusterFeatures,
                                   c("Estimate",
                                     "lower",
                                     "OR",
                                     "upper",
                                     "full.AUC",
                                     "Delta.AUC",
                                     "z.IDI",
                                     "Frequency")]

nugget <- fc$membership
names(nugget) <- clusterFeatures

tableMCI_De$Cluster <- nugget[rownames(tableMCI_De)]

rnames <- clusterFeatures[str_detect(clusterFeatures,"ST")]
frnames <- rnames
rnames <- str_replace_all(rnames,"M_","")
rnames <- str_replace_all(rnames,"RD_","")
description <- character()

for (ddet in c(1:length(rnames)))
{
  description <- c(description,TADPOLE_D1_D2_Dict$TEXT[str_detect(TADPOLE_D1_D2_Dict$FLDNAME,rnames[ddet])][1])
}
names(description) <- frnames

tableMCI_De$Description <- description[rownames(tableMCI_De)]
pander::pander(tableMCI_De)

	Estimate	lower	OR	upper	full.AUC	Delta.AUC	z.IDI	Frequency	Cluster	Description
M_ST32CV	-14.298	2.01e-08	6.17e-07	1.90e-05	0.739	4.00e-02	8.11	1.00	1	Volume (Cortical Parcellation) of LeftInferiorTemporal
M_ST24TA	-24.833	6.09e-15	1.64e-11	4.42e-08	0.739	2.86e-02	7.02	1.00	1	Cortical Thickness Average of LeftEntorhinal
M_ST46TS	107.131	2.05e+18	3.36e+46	5.51e+74	0.737	1.25e-02	5.36	1.00	1	Cortical Thickness Standard Deviation of LeftParsOrbitalis
M_ST30SV	2.881	8.13e+00	1.78e+01	3.91e+01	0.705	2.35e-02	7.72	0.90	2	Volume (WM Parcellation) of LeftInferiorLateralVentricle
M_ST32SA	-2.990	9.67e-03	5.03e-02	2.61e-01	0.727	1.96e-02	5.64	0.90	2	Surface Area of LeftInferiorTemporal
Hippocampus	-13.711	5.93e-08	1.11e-06	2.08e-05	0.730	4.61e-02	8.64	1.00	3	NA
M_ST12SV	-10.998	5.36e-07	1.67e-05	5.23e-04	0.722	2.87e-02	7.36	0.90	2	Volume (WM Parcellation) of LeftAmygdala
M_ST31CV	-9.623	3.00e-06	6.62e-05	1.46e-03	0.730	3.22e-02	7.02	0.90	3	Volume (Cortical Parcellation) of LeftInferiorParietal
M_ST60TS	65.901	2.55e+12	4.17e+28	6.83e+44	0.733	2.36e-02	5.31	0.95	3	Cortical Thickness Standard Deviation of LeftTemporalPole
RD_ST31TA	0.846	1.49e+00	2.33e+00	3.65e+00	0.702	1.13e-02	5.54	0.55	2	Cortical Thickness Average of LeftInferiorParietal
M_ST40CV	-10.650	9.62e-07	2.37e-05	5.84e-04	0.718	2.55e-02	7.10	0.95	4	Volume (Cortical Parcellation) of LeftMiddleTemporal
M_ST39CV	9.367	5.56e+01	1.17e+04	2.46e+06	0.734	1.76e-02	5.54	0.65	2	Volume (Cortical Parcellation) of LeftMedialOrbitofrontal
M_ST24CV	-20.188	3.20e-11	1.71e-09	9.12e-08	0.724	3.97e-02	8.64	0.95	4	Volume (Cortical Parcellation) of LeftEntorhinal
RD_ST52TA	1.696	1.76e+00	5.45e+00	1.69e+01	0.720	-3.62e-05	4.74	0.75	2	Cortical Thickness Average of LeftPrecuneus
M_ST43TS	131.338	9.38e+24	1.09e+57	1.28e+89	0.720	1.73e-02	5.33	0.90	4	Cortical Thickness Standard Deviation of LeftParacentral
M_ST35SA	-3.534	3.29e-03	2.92e-02	2.59e-01	0.736	2.53e-02	5.26	0.95	3	Surface Area of LeftLateralOccipital
M_ST129TS	79.463	8.02e+17	3.24e+34	1.31e+51	0.715	1.35e-02	5.47	0.70	5	Cortical Thickness Standard Deviation of LeftInsula
M_ST24SA	2.887	3.23e+00	1.79e+01	9.97e+01	0.724	1.54e-02	5.51	0.25	4	Surface Area of LeftEntorhinal
M_ST31TA	-70.265	1.23e-36	3.05e-31	7.54e-26	0.731	5.00e-02	9.09	0.50	2	Cortical Thickness Average of LeftInferiorParietal
M_ST26CV	-4.445	1.04e-03	1.17e-02	1.32e-01	0.715	1.40e-02	5.02	0.55	5	Volume (Cortical Parcellation) of LeftFusiform
M_ST40TA	-79.016	1.07e-40	4.83e-35	2.18e-29	0.718	8.71e-02	11.13	0.65	5	Cortical Thickness Average of LeftMiddleTemporal
RD_ST44TA	0.503	1.18e+00	1.65e+00	2.32e+00	0.712	1.28e-02	4.93	0.50	5	Cortical Thickness Average of LeftParahippocampal
M_ST45TA	32.731	5.97e+07	1.64e+14	4.51e+20	0.708	2.94e-02	6.40	0.40	5	Cortical Thickness Average of LeftParsOpercularis
M_ST32TA	-30.317	3.08e-16	6.81e-14	1.51e-11	0.705	6.70e-02	10.34	0.40	5	Cortical Thickness Average of LeftInferiorTemporal
M_ST44TS	-53.998	1.21e-41	3.54e-24	1.03e-06	0.717	1.44e-02	4.47	0.55	5	Cortical Thickness Standard Deviation of LeftParahippocampal
M_ST43TA	20.526	2.83e+02	8.21e+08	2.38e+15	0.715	2.10e-02	4.78	0.25	2	Cortical Thickness Average of LeftParacentral
M_ST55SA	-0.886	2.76e-01	4.12e-01	6.16e-01	0.696	1.37e-02	5.86	0.30	5	Surface Area of LeftRostralMiddleFrontal
M_ST60TA	-5.852	4.15e-04	2.87e-03	1.99e-02	0.704	3.23e-02	6.49	0.25	6	Cortical Thickness Average of LeftTemporalPole
M_ST55TS	115.895	1.13e+29	2.15e+50	4.08e+71	0.715	3.17e-02	6.19	0.40	5	Cortical Thickness Standard Deviation of LeftRostralMiddleFrontal

3.1 Decorrelating the sets

TADPOLE_DX_TRAIND <- GDSTMDecorrelation(TADPOLE_DX_TRAIN,Outcome="DX",
                                        thr=0.6,
                                        type="RLM",
                                        method="spearman",
                                        verbose = TRUE)
#> 
#>  Included: 200 , Uni p: 0.0118249 To Outcome: 93 , Base: 5 , In Included: 5 , Base Cor: 14 
#> 1 , Top: 40 < 0.6 >( 3 )[ 1 : 0 : 0.594 ]( 39 , 89 , 0 ),<>Tot Used: 128 , Added: 89 , Zero Std: 0 , Max Cor: 0.9882019 
#> 2 , Top: 26 < 0.6 >[ FALSE ]( 1 )[ 1 : 0 : 0.594 ]( 22 , 32 , 39 ),<>Tot Used: 147 , Added: 32 , Zero Std: 0 , Max Cor: 0.9458602 
#> 3 , Top: 16 < 0.6 >( 1 )[ 1 : 0 : 0.594 ]( 15 , 18 , 58 ),<>Tot Used: 154 , Added: 18 , Zero Std: 0 , Max Cor: 0.8804423 
#> 4 , Top: 12 < 0.6 >( 2 )[ 1 : 0 : 0.594 ]( 9 , 10 , 68 ),<>Tot Used: 156 , Added: 10 , Zero Std: 0 , Max Cor: 0.891374 
#> 5 , Top: 8 < 0.6 >( 1 )[ 1 : 0 : 0 ]( 8 , 11 , 75 ),<>Tot Used: 158 , Added: 11 , Zero Std: 0 , Max Cor: 0.8757232 
#> 6 , Top: 4 < 0.6 >[ TRUE ]( 1 )[ 1 : 0 : 0 ]( 4 , 4 , 78 ),<>Tot Used: 158 , Added: 4 , Zero Std: 0 , Max Cor: 0.6082068 
#> 7 , Top: 1 < 0.6 >[ FALSE ]( 1 )[ 1 : 0 : 0.6 ]( 1 , 1 , 82 ),<>Tot Used: 158 , Added: 1 , Zero Std: 0 , Max Cor: 0.5990138 
#> [ 8 ], 0.5979913 . Cor to Base: 102 , ABase: 76
TADPOLE_DX_TESTD <-  predictDecorrelate(TADPOLE_DX_TRAIND,TADPOLE_DX_TEST)

3.1.1 Decorrelated ML

bDXmlD <- BSWiMS.model(DX~.,TADPOLE_DX_TRAIND,NumberofRepeats = 20)

[+++++-+++-+++-+++–+++++-++++-++–++++-++++-+++-+-+++++-++++-++++–+++—++++–+++–++++-++++-+++-+++-]…….

pander::pander(bDXmlD$bagging$Jaccard.SM)

0.14

fs <- bDXmlD$bagging$frequencyTable
barplot(fs[order(-fs)],las=2,main="Selected Features",cex.names = 0.5)

sm <- summary(bDXmlD)
pander::pander(sm$coefficients)

	Estimate	lower	OR	upper	u.Accuracy	r.Accuracy	full.Accuracy	u.AUC	r.AUC	full.AUC	IDI	NRI	z.IDI	z.NRI	Delta.AUC	Frequency
Ba_M_ST24TA	-82.8668	8.56e-42	1.03e-36	1.23e-31	0.693	0.672	0.729	0.682	0.659	0.725	0.1410	0.728	11.78	11.44	0.06618	1.00
Ba_M_ST59TA	-165.5492	4.89e-85	1.27e-72	3.29e-60	0.609	0.665	0.731	0.605	0.659	0.729	0.1186	0.567	10.62	8.69	0.07078	1.00
M_ST49CV	-9.7934	5.11e-06	5.58e-05	6.10e-04	0.594	0.661	0.698	0.597	0.644	0.700	0.0828	0.489	8.66	7.41	0.05612	0.65
M_ST35CV	-12.4938	1.05e-07	3.75e-06	1.34e-04	0.631	0.682	0.714	0.636	0.677	0.716	0.0650	0.462	7.67	6.98	0.03935	1.00
De_M_ST30SV	8.5509	2.92e+02	5.17e+03	9.15e+04	0.632	0.695	0.721	0.630	0.695	0.721	0.0568	0.483	7.15	7.30	0.02657	1.00
De_M_ST49TA	34.0464	2.74e+08	6.11e+14	1.36e+21	0.543	0.665	0.698	0.554	0.657	0.698	0.0460	0.369	6.48	5.50	0.04106	0.20
De_M_ST32TA	-67.7685	6.08e-43	3.70e-30	2.26e-17	0.568	0.699	0.734	0.573	0.699	0.734	0.0409	0.355	6.37	5.29	0.03480	0.45
Ba_Ventricles	0.1215	1.07e+00	1.13e+00	1.19e+00	0.617	0.668	0.686	0.619	0.662	0.685	0.0389	0.381	5.67	5.70	0.02302	0.20
Ba_M_ST35SA	-4.7798	8.65e-04	8.40e-03	8.15e-02	0.565	0.717	0.732	0.580	0.711	0.730	0.0356	0.440	5.65	6.62	0.01823	0.65
De_M_ST62TA	44.4839	1.42e+08	2.08e+19	3.07e+30	0.580	0.691	0.716	0.584	0.691	0.715	0.0358	0.412	5.65	6.16	0.02349	0.95
De_M_ST39TA	-3.9097	2.97e-03	2.00e-02	1.35e-01	0.584	0.669	0.688	0.593	0.660	0.683	0.0379	0.318	5.59	4.73	0.02295	0.20
De_M_ST23TA	13.8805	3.78e+02	1.07e+06	3.01e+09	0.570	0.684	0.694	0.564	0.683	0.696	0.0362	0.304	5.58	4.50	0.01340	0.10
Ba_M_ST46TS	188.6648	1.26e+33	8.63e+81	5.89e+130	0.588	0.715	0.736	0.563	0.716	0.735	0.0367	0.336	5.56	5.00	0.01879	0.65
RD_ST57TA	0.2686	1.09e+00	1.31e+00	1.56e+00	0.624	0.696	0.723	0.588	0.684	0.711	0.0363	0.370	5.44	5.60	0.02640	0.10
Ba_M_ST129TS	12.1062	7.31e+01	1.81e+05	4.48e+08	0.605	0.671	0.690	0.578	0.673	0.699	0.0343	0.317	5.41	4.71	0.02510	0.10
Ba_M_ST21SV	3.1103	3.51e+00	2.24e+01	1.43e+02	0.598	0.668	0.689	0.603	0.668	0.693	0.0325	0.239	5.38	3.52	0.02485	0.45
De_M_ST55SA	-1.1207	1.53e-01	3.26e-01	6.95e-01	0.597	0.708	0.723	0.598	0.703	0.723	0.0339	0.409	5.28	6.11	0.01968	0.30
De_M_ST26TA	-23.6746	2.26e-17	5.23e-11	1.21e-04	0.550	0.707	0.732	0.544	0.702	0.730	0.0302	0.269	5.27	3.97	0.02740	0.15
Ba_M_ST32SA	-7.0635	1.18e-05	8.56e-04	6.19e-02	0.619	0.724	0.730	0.622	0.721	0.726	0.0339	0.383	5.24	5.71	0.00523	1.00
De_M_ST45TA	161.7732	9.47e+28	1.81e+70	3.45e+111	0.600	0.724	0.735	0.593	0.720	0.730	0.0323	0.405	5.20	6.06	0.01010	0.95
RD_ST44TA	1.5837	1.73e+00	4.87e+00	1.37e+01	0.606	0.722	0.735	0.573	0.722	0.732	0.0313	0.332	5.15	4.99	0.01073	0.65
RD_ST29SV	0.1569	1.06e+00	1.17e+00	1.30e+00	0.635	0.676	0.683	0.589	0.679	0.692	0.0292	0.375	5.14	5.70	0.01282	0.10
M_ST60TS	76.5383	1.12e+13	1.74e+33	2.71e+53	0.595	0.694	0.711	0.592	0.693	0.709	0.0292	0.345	5.11	5.12	0.01634	1.00
RD_ST55TA	1.2573	1.55e+00	3.52e+00	7.96e+00	0.606	0.718	0.742	0.561	0.716	0.740	0.0271	0.328	5.02	4.95	0.02391	0.20
De_M_ST11SV	-110.0996	2.02e-70	1.53e-48	1.16e-26	0.634	0.709	0.721	0.624	0.710	0.719	0.0310	0.310	4.92	4.61	0.00926	0.95
RD_ST24TA	0.0136	1.00e+00	1.01e+00	1.02e+00	0.621	0.659	0.680	0.572	0.662	0.676	0.0244	0.169	4.68	2.53	0.01358	0.10
De_M_ST31SA	-1.6079	6.59e-02	2.00e-01	6.09e-01	0.557	0.720	0.744	0.566	0.718	0.742	0.0248	0.276	4.66	4.08	0.02336	0.25
RD_ST52TA	4.0332	3.45e+00	5.64e+01	9.25e+02	0.616	0.715	0.726	0.557	0.718	0.722	0.0275	0.193	4.64	2.91	0.00335	0.90
De_M_ST43TA	87.0517	1.81e+12	6.40e+37	2.27e+63	0.574	0.716	0.725	0.577	0.713	0.724	0.0267	0.353	4.57	5.25	0.01109	0.90
RD_ST31TA	0.2559	1.09e+00	1.29e+00	1.54e+00	0.627	0.679	0.701	0.575	0.681	0.694	0.0210	0.156	4.05	2.35	0.01306	0.25
Ba_M_ST43TS	6.9809	1.59e+00	1.08e+03	7.28e+05	0.574	0.671	0.690	0.551	0.669	0.688	0.0198	0.341	3.85	5.07	0.01881	0.10
De_RD_ST52CV	2.3747	1.44e+00	1.07e+01	8.01e+01	0.583	0.684	0.697	0.560	0.688	0.698	0.0153	0.157	3.60	2.32	0.00993	0.75
De_M_ST29SV	125.1641	2.56e+01	2.28e+54	2.03e+107	0.469	0.723	0.730	0.555	0.720	0.727	0.0142	0.555	3.53	8.99	0.00698	0.95

pander::pander(bDXmlD$univariate[bDXmlD$selectedfeatures,])

	Name	RName	ZUni
Ba_M_ST24TA	Ba_M_ST24TA	Ba_M_ST24TA	9.11
Ba_M_ST32SA	Ba_M_ST32SA	Ba_M_ST32SA	6.46
RD_ST55TA	RD_ST55TA	RD_ST55TA	3.01
RD_ST52TA	RD_ST52TA	RD_ST52TA	4.65
De_M_ST30SV	De_M_ST30SV	De_M_ST30SV	6.64
De_M_ST43TA	De_M_ST43TA	De_M_ST43TA	3.71
De_M_ST45TA	De_M_ST45TA	De_M_ST45TA	3.77
De_M_ST11SV	De_M_ST11SV	De_M_ST11SV	6.41
De_M_ST129CV	De_M_ST129CV	De_M_ST129CV	2.62
De_M_ST29SV	De_M_ST29SV	De_M_ST29SV	1.97
M_ST35CV	M_ST35CV	M_ST35CV	6.11
RD_ST57TA	RD_ST57TA	RD_ST57TA	3.88
Ba_M_ST59TA	Ba_M_ST59TA	Ba_M_ST59TA	6.44
De_M_ST62TA	De_M_ST62TA	De_M_ST62TA	3.50
M_ST60TS	M_ST60TS	M_ST60TS	4.29
Ba_M_ST35SA	Ba_M_ST35SA	Ba_M_ST35SA	3.29
Ba_RD_ST52SA	Ba_RD_ST52SA	Ba_RD_ST52SA	1.29
De_M_ST23TA	De_M_ST23TA	De_M_ST23TA	2.73
Ba_M_ST46TS	Ba_M_ST46TS	Ba_M_ST46TS	2.82
RD_ST44TA	RD_ST44TA	RD_ST44TA	3.73
De_M_ST55SA	De_M_ST55SA	De_M_ST55SA	4.37
De_M_ST31SA	De_M_ST31SA	De_M_ST31SA	3.88
M_ST49CV	M_ST49CV	M_ST49CV	5.26
De_M_ST26TA	De_M_ST26TA	De_M_ST26TA	3.05
De_M_ST32TA	De_M_ST32TA	De_M_ST32TA	3.12
De_RD_ST52CV	De_RD_ST52CV	De_RD_ST52CV	3.97
Ba_M_ST21SV	Ba_M_ST21SV	Ba_M_ST21SV	5.38
RD_ST29SV	RD_ST29SV	RD_ST29SV	3.65
Ba_M_ST129TS	Ba_M_ST129TS	Ba_M_ST129TS	3.90
Ba_Ventricles	Ba_Ventricles	Ba_Ventricles	5.44
De_M_ST39TA	De_M_ST39TA	De_M_ST39TA	4.66
De_M_ST49TA	De_M_ST49TA	De_M_ST49TA	1.40
RD_ST24TA	RD_ST24TA	RD_ST24TA	4.32
Ba_M_ST43TS	Ba_M_ST43TS	Ba_M_ST43TS	2.94
RD_ST31TA	RD_ST31TA	RD_ST31TA	4.55
De_M_ST47TA	De_M_ST47TA	De_M_ST47TA	3.19
De_M_ST59CV	De_M_ST59CV	De_M_ST59CV	3.73
De_M_ST36TA	De_M_ST36TA	De_M_ST36TA	2.52

prBin <- predictionStats_binary(cbind(TADPOLE_DX_TESTD$DX,predict(bDXmlD,TADPOLE_DX_TESTD)),"MCI vs Dementia")

MCI vs Dementia

pander::pander(prBin$aucs)

est	lower	upper
0.804	0.755	0.853

pander::pander(prBin$accc)

est	lower	upper
0.736	0.689	0.78

pander::pander(prBin$berror)

50%	2.5%	97.5%
0.27	0.22	0.326

pander::pander(prBin$sensitivity)

est	lower	upper
0.714	0.614	0.801

par(op)

3.1.2 The formula network

cmax <- apply(bDXmlD$bagging$formulaNetwork,2,max)
cnames <- names(cmax[cmax>=0.25])
cmax <- cmax[cnames]

adma <- bDXmlD$bagging$formulaNetwork[cnames,cnames]

adma[adma<0.15] <- 0;
gr <- graph_from_adjacency_matrix(adma,mode = "undirected",diag = FALSE,weighted=TRUE)
gr$layout <- layout_with_fr

fc <- cluster_optimal(gr)
plot(fc, gr,
     edge.width=10*E(gr)$weight,
     vertex.size=10*cmax,
     vertex.label.cex=0.75,
     vertex.label.dist=0,
     main="MCI vs Dementia Diagnosis")

par(op)

3.1.3 Decorrelated MCI vs AD table

clusterFeatures <- fc$names

tableMCI_DeD <- sm$coefficients[clusterFeatures,
                                   c("Estimate",
                                     "lower",
                                     "OR",
                                     "upper",
                                     "full.AUC",
                                     "Delta.AUC",
                                     "z.IDI",
                                     "Frequency")]

nugget <- fc$membership
names(nugget) <- clusterFeatures

tableMCI_DeD$Cluster <- nugget[rownames(tableMCI_DeD)]

rnames <- clusterFeatures[str_detect(clusterFeatures,"ST")]
frnames <- rnames
rnames <- str_replace_all(rnames,"M_","")
rnames <- str_replace_all(rnames,"RD_","")
rnames <- str_replace_all(rnames,"Ba_","")
rnames <- str_replace_all(rnames,"De_","")
description <- character()

for (ddet in c(1:length(rnames)))
{
  description <- c(description,TADPOLE_D1_D2_Dict$TEXT[str_detect(TADPOLE_D1_D2_Dict$FLDNAME,rnames[ddet])][1])
}
names(description) <- frnames

tableMCI_DeD$Description <- description[rownames(tableMCI_DeD)]



## Getting the decorrelation formula
dc <- getDerivedCoefficients(TADPOLE_DX_TRAIND)
decornames <- rownames(sm$coefficients)

deNames_in_dc <- decornames[decornames %in% names(dc)]
theDeFormulas <- dc[deNames_in_dc]
deFromula <- character(length(theDeFormulas))
names(deFromula) <- names(theDeFormulas)
for (dx in names(deFromula))
{
  coef <- theDeFormulas[[dx]]
  cname <- names(theDeFormulas[[dx]])
  names(cname) <- cname
  for (cf in names(coef))
  {
    if (cf != dx)
    {
      if (coef[cf]>0)
      {
        deFromula[dx] <- paste(deFromula[dx],
                               sprintf("+ %5.3f*%s",coef[cf],cname[cf]))
      }
      else
      {
        deFromula[dx] <- paste(deFromula[dx],
                               sprintf("%5.3f*%s",coef[cf],cname[cf]))
      }
    }
  }
}

tableMCI_DeD$DecorFormula <- deFromula[rownames(tableMCI_DeD)]



pander::pander(tableMCI_DeD)

	Estimate	lower	OR	upper	full.AUC	Delta.AUC	z.IDI	Frequency	Cluster	Description	DecorFormula
Ba_M_ST24TA	-82.867	8.56e-42	1.03e-36	1.23e-31	0.725	0.06618	11.78	1.00	1	Cortical Thickness Average of LeftEntorhinal	NA
Ba_M_ST32SA	-7.063	1.18e-05	8.56e-04	6.19e-02	0.726	0.00523	5.24	1.00	1	Surface Area of LeftInferiorTemporal	NA
RD_ST52TA	4.033	3.45e+00	5.64e+01	9.25e+02	0.722	0.00335	4.64	0.90	1	Cortical Thickness Average of LeftPrecuneus	NA
De_M_ST30SV	8.551	2.92e+02	5.17e+03	9.15e+04	0.721	0.02657	7.15	1.00	2	Volume (WM Parcellation) of LeftInferiorLateralVentricle	-0.331Ventricles + 1.000M_ST30SV
De_M_ST43TA	87.052	1.81e+12	6.40e+37	2.27e+63	0.724	0.01109	4.57	0.90	1	Cortical Thickness Average of LeftParacentral	+ 1.000M_ST43TA -0.792M_ST59TA
De_M_ST45TA	161.773	9.47e+28	1.81e+70	3.45e+111	0.730	0.01010	5.20	0.95	1	Cortical Thickness Average of LeftParsOpercularis	+ 1.000M_ST45TA -0.733M_ST59TA
De_M_ST11SV	-110.100	2.02e-70	1.53e-48	1.16e-26	0.719	0.00926	4.92	0.95	2	Volume (WM Parcellation) of LeftAccumbensArea	+ 0.000ICV + 1.000M_ST11SV
De_M_ST29SV	125.164	2.56e+01	2.28e+54	2.03e+107	0.727	0.00698	3.53	0.95	2	Volume (WM Parcellation) of LeftHippocampus	-0.794Hippocampus -0.007M_ST24TA + 1.000*M_ST29SV
M_ST35CV	-12.494	1.05e-07	3.75e-06	1.34e-04	0.716	0.03935	7.67	1.00	2	Volume (Cortical Parcellation) of LeftLateralOccipital	NA
Ba_M_ST59TA	-165.549	4.89e-85	1.27e-72	3.29e-60	0.729	0.07078	10.62	1.00	3	Cortical Thickness Average of LeftSupramarginal	NA
De_M_ST62TA	44.484	1.42e+08	2.08e+19	3.07e+30	0.715	0.02349	5.65	0.95	2	Cortical Thickness Average of LeftTransverseTemporal	-0.910M_ST59TA + 1.000M_ST62TA
M_ST60TS	76.538	1.12e+13	1.74e+33	2.71e+53	0.709	0.01634	5.11	1.00	2	Cortical Thickness Standard Deviation of LeftTemporalPole	NA
Ba_M_ST35SA	-4.780	8.65e-04	8.40e-03	8.15e-02	0.730	0.01823	5.65	0.65	3	Surface Area of LeftLateralOccipital	NA
Ba_M_ST46TS	188.665	1.26e+33	8.63e+81	5.89e+130	0.735	0.01879	5.56	0.65	3	Cortical Thickness Standard Deviation of LeftParsOrbitalis	NA
RD_ST44TA	1.584	1.73e+00	4.87e+00	1.37e+01	0.732	0.01073	5.15	0.65	3	Cortical Thickness Average of LeftParahippocampal	NA
De_M_ST55SA	-1.121	1.53e-01	3.26e-01	6.95e-01	0.723	0.01968	5.28	0.30	3	Surface Area of LeftRostralMiddleFrontal	NA
De_M_ST31SA	-1.608	6.59e-02	2.00e-01	6.09e-01	0.742	0.02336	4.66	0.25	3	Surface Area of LeftInferiorParietal	NA
M_ST49CV	-9.793	5.11e-06	5.58e-05	6.10e-04	0.700	0.05612	8.66	0.65	4	Volume (Cortical Parcellation) of LeftPostcentral	NA
De_M_ST32TA	-67.768	6.08e-43	3.70e-30	2.26e-17	0.734	0.03480	6.37	0.45	3	Cortical Thickness Average of LeftInferiorTemporal	+ 1.000M_ST32TA -0.841M_ST59TA
De_RD_ST52CV	2.375	1.44e+00	1.07e+01	8.01e+01	0.698	0.00993	3.60	0.75	4	Volume (Cortical Parcellation) of LeftPrecuneus	-0.472RD_ST52SA + 1.000RD_ST52CV
Ba_M_ST21SV	3.110	3.51e+00	2.24e+01	1.43e+02	0.693	0.02485	5.38	0.45	4	Volume (WM Parcellation) of LeftChoroidPlexus	NA
RD_ST31TA	0.256	1.09e+00	1.29e+00	1.54e+00	0.694	0.01306	4.05	0.25	5	Cortical Thickness Average of LeftInferiorParietal	NA

4 Diagnosis NL vs AD

4.0.1 the set

TrainFraction <- 0.60;

table(validBaselineTadpole$DX)

Dementia MCI NL 244 711 452

TADPOLECrossMRI <- subset(validBaselineTadpole,DX == "Dementia" | DX == "NL")
table(TADPOLECrossMRI$DX)

Dementia MCI NL 244 0 452

TADPOLECrossMRI$DX <- 1*(as.character(TADPOLECrossMRI$DX) == "Dementia")
table(TADPOLECrossMRI$DX)

0 1 452 244

TADPOLECrossMRI$ADAS13 <- NULL
TADPOLECrossMRI$ADAS11 <- NULL
TADPOLECrossMRI$MMSE <- NULL
TADPOLECrossMRI$RAVLT_immediate <- NULL
TADPOLECrossMRI$RAVLT_learning <- NULL
TADPOLECrossMRI$RAVLT_perc_forgetting <- NULL
TADPOLECrossMRI$RAVLT_forgetting <- NULL
TADPOLECrossMRI$FAQ <- NULL

TADPOLE_Cases <- subset(TADPOLECrossMRI,DX==1)
TADPOLE_Controls <- subset(TADPOLECrossMRI,DX==0)
trainCasesSet <- sample(nrow(TADPOLE_Cases),nrow(TADPOLE_Cases)*TrainFraction)
trainControlSet <- sample(nrow(TADPOLE_Controls),nrow(TADPOLE_Controls)*TrainFraction)

TADPOLE_DX_NLDE_TRAIN <- rbind(TADPOLE_Cases[trainCasesSet,],TADPOLE_Controls[trainControlSet,])
TADPOLE_DX_NLDE_TEST <- TADPOLECrossMRI[!(rownames(TADPOLECrossMRI) %in% rownames(TADPOLE_DX_NLDE_TRAIN)),]


pander::pander(table(TADPOLE_DX_NLDE_TRAIN$DX))

0	1
271	146

pander::pander(table(TADPOLE_DX_NLDE_TEST$DX))

0	1
181	98

par(op)

4.0.2 Learning

bDXmlNLDE <- BSWiMS.model(DX~.,TADPOLE_DX_NLDE_TRAIN,NumberofRepeats = 20)

[++-+++-+++-++–+++-++-+++-++-+++-+++-++-++-++–+++-++-++—+++-++–++-++–]….

pander::pander(bDXmlNLDE$bagging$Jaccard.SM)

0.195

fs <- bDXmlNLDE$bagging$frequencyTable
barplot(fs[order(-fs)],las=2,main="Selected Features",cex.names = 0.5)

sm <- summary(bDXmlNLDE)
pander::pander(sm$coefficients)

	Estimate	lower	OR	upper	u.Accuracy	r.Accuracy	full.Accuracy	u.AUC	r.AUC	full.AUC	IDI	NRI	z.IDI	z.NRI	Delta.AUC	Frequency
M_ST24CV	-126.03	5.06e-66	1.84e-55	6.68e-45	0.836	0.851	0.900	0.829	0.848	0.896	0.1253	1.007	9.09	13.82	0.04885	1.00
M_ST51TA	324.67	9.53e+92	1.01e+141	1.07e+189	0.607	0.886	0.918	0.600	0.883	0.916	0.0831	0.824	7.30	10.78	0.03333	0.85
M_ST24TA	-153.90	6.39e-92	1.45e-67	3.29e-43	0.851	0.892	0.917	0.843	0.891	0.915	0.0769	0.790	6.96	10.40	0.02455	1.00
M_ST31TA	-212.64	5.58e-133	4.47e-93	3.58e-53	0.745	0.891	0.916	0.737	0.887	0.914	0.0757	0.666	6.89	8.71	0.02753	0.55
M_ST40TA	-207.63	4.70e-123	6.70e-91	9.57e-59	0.787	0.865	0.891	0.780	0.864	0.888	0.0756	0.803	6.72	10.72	0.02418	0.50
Hippocampus	-58.83	1.17e-38	2.83e-26	6.84e-14	0.840	0.896	0.917	0.839	0.893	0.915	0.0577	0.754	5.84	9.72	0.02196	1.00
M_ST45TA	162.32	1.76e+44	3.13e+70	5.57e+96	0.660	0.866	0.890	0.657	0.867	0.889	0.0574	0.785	5.79	10.16	0.02181	0.35
M_ST30SV	9.76	6.24e+01	1.73e+04	4.79e+06	0.781	0.874	0.891	0.778	0.872	0.888	0.0475	0.640	5.42	8.17	0.01592	0.40
M_ST12SV	-99.35	3.37e-62	7.11e-44	1.50e-25	0.802	0.883	0.900	0.796	0.879	0.897	0.0521	0.622	5.39	7.94	0.01839	0.95
M_ST36CV	41.51	5.35e+06	1.07e+18	2.12e+29	0.600	0.885	0.902	0.595	0.881	0.898	0.0412	0.577	5.12	7.11	0.01678	0.55
M_ST39CV	31.34	1.34e+05	4.09e+13	1.25e+22	0.595	0.880	0.897	0.594	0.876	0.894	0.0416	0.395	5.02	4.75	0.01809	0.60
M_ST31CV	-21.05	4.55e-16	7.24e-10	1.15e-03	0.746	0.895	0.914	0.735	0.891	0.912	0.0417	0.661	4.91	8.27	0.02055	0.35
M_ST32TA	-59.88	6.78e-40	9.85e-27	1.43e-13	0.810	0.877	0.893	0.801	0.872	0.889	0.0380	0.548	4.80	6.79	0.01766	0.40
M_ST40CV	-33.71	2.01e-25	2.29e-15	2.60e-05	0.763	0.889	0.906	0.762	0.885	0.903	0.0398	0.567	4.79	7.12	0.01798	0.65
M_ST15TA	-111.39	6.38e-84	4.20e-49	2.77e-14	0.708	0.898	0.911	0.703	0.896	0.911	0.0342	0.448	4.46	5.43	0.01467	0.30
M_ST32CV	-30.61	2.31e-23	5.09e-14	1.12e-04	0.769	0.876	0.890	0.765	0.874	0.886	0.0402	0.712	4.38	9.14	0.01251	0.60
M_ST32SA	-1.63	4.63e-02	1.96e-01	8.30e-01	0.672	0.865	0.880	0.673	0.863	0.881	0.0311	0.534	4.24	6.52	0.01734	0.10
M_ST129CV	7.45	5.89e+00	1.73e+03	5.07e+05	0.595	0.889	0.901	0.602	0.883	0.898	0.0311	0.357	4.24	4.24	0.01536	0.10
M_ST24SA	17.83	1.47e+01	5.56e+07	2.10e+14	0.588	0.892	0.904	0.595	0.888	0.900	0.0287	0.572	4.14	7.04	0.01248	0.45
M_ST13CV	-7.36	1.05e-06	6.37e-04	3.87e-01	0.723	0.875	0.890	0.727	0.874	0.888	0.0262	0.739	4.09	9.43	0.01450	0.15
RD_ST57TA	5.10	2.67e+00	1.64e+02	1.00e+04	0.635	0.888	0.900	0.604	0.883	0.897	0.0260	0.330	4.01	3.97	0.01429	0.45
RD_ST29SV	3.90	2.38e+00	4.94e+01	1.02e+03	0.674	0.873	0.888	0.655	0.869	0.887	0.0319	0.749	3.99	9.60	0.01862	0.25
RD_ST32TA	4.64	1.49e+00	1.04e+02	7.25e+03	0.683	0.908	0.911	0.656	0.906	0.909	0.0282	0.585	3.56	7.36	0.00294	0.80
RD_ST55TA	3.07	1.38e+00	2.15e+01	3.35e+02	0.608	0.893	0.904	0.585	0.889	0.902	0.0206	0.417	3.46	5.08	0.01329	0.25
M_ST15CV	-6.86	1.15e-06	1.05e-03	9.54e-01	0.643	0.890	0.892	0.644	0.886	0.889	0.0189	0.448	2.94	5.57	0.00217	0.20

pander::pander(bDXmlNLDE$univariate[bDXmlNLDE$selectedfeatures,])

	Name	RName	ZUni
M_ST24TA	M_ST24TA	M_ST24TA	20.33
Hippocampus	Hippocampus	Hippocampus	19.17
RD_ST32TA	RD_ST32TA	RD_ST32TA	7.38
M_ST51TA	M_ST51TA	M_ST51TA	5.93
M_ST13SA	M_ST13SA	M_ST13SA	4.55
M_ST31TA	M_ST31TA	M_ST31TA	12.01
M_ST24CV	M_ST24CV	M_ST24CV	18.01
ST3SV	ST3SV	ST3SV	6.52
M_ST24SA	M_ST24SA	M_ST24SA	4.64
M_ST40CV	M_ST40CV	M_ST40CV	14.07
M_ST31CV	M_ST31CV	M_ST31CV	11.24
M_ST12SV	M_ST12SV	M_ST12SV	16.05
M_ST15TA	M_ST15TA	M_ST15TA	10.18
M_ST32TA	M_ST32TA	M_ST32TA	14.58
M_ST129TA	M_ST129TA	M_ST129TA	7.11
M_ST54CV	M_ST54CV	M_ST54CV	3.15
RD_ST57TA	RD_ST57TA	RD_ST57TA	5.39
M_ST36CV	M_ST36CV	M_ST36CV	5.52
M_ST32CV	M_ST32CV	M_ST32CV	14.59
M_ST39CV	M_ST39CV	M_ST39CV	4.65
M_ST129CV	M_ST129CV	M_ST129CV	5.40
M_ST30SV	M_ST30SV	M_ST30SV	13.99
M_ST60SA	M_ST60SA	M_ST60SA	2.93
M_ST40TA	M_ST40TA	M_ST40TA	14.19
RD_ST55TA	RD_ST55TA	RD_ST55TA	3.46
RD_ST29SV	RD_ST29SV	RD_ST29SV	6.99
M_ST24TS	M_ST24TS	M_ST24TS	5.43
M_ST32SA	M_ST32SA	M_ST32SA	7.95
M_ST45TA	M_ST45TA	M_ST45TA	6.90
M_ST15CV	M_ST15CV	M_ST15CV	7.45
ST127SV	ST127SV	ST127SV	5.88
M_ST32TS	M_ST32TS	M_ST32TS	1.66
M_ST23CV	M_ST23CV	M_ST23CV	2.51
M_ST52CV	M_ST52CV	M_ST52CV	9.63
M_ST13CV	M_ST13CV	M_ST13CV	10.07

prBin <- predictionStats_binary(cbind(TADPOLE_DX_NLDE_TEST$DX,predict(bDXmlNLDE,TADPOLE_DX_NLDE_TEST)),"NL vs Dementia")

NL vs Dementia

pander::pander(prBin$aucs)

est	lower	upper
0.938	0.907	0.97

pander::pander(prBin$accc)

est	lower	upper
0.864	0.818	0.902

pander::pander(prBin$berror)

50%	2.5%	97.5%
0.146	0.101	0.19

pander::pander(prBin$sensitivity)

est	lower	upper
0.816	0.725	0.887

par(op)

4.0.3 The formula network

cmax <- apply(bDXmlNLDE$bagging$formulaNetwork,2,max)
cnames <- names(cmax[cmax>=0.25])
cmax <- cmax[cnames]

adma <- bDXmlNLDE$bagging$formulaNetwork[cnames,cnames]

adma[adma<0.15] <- 0;
gr <- graph_from_adjacency_matrix(adma,mode = "undirected",diag = FALSE,weighted=TRUE)
gr$layout <- layout_with_fr

fc <- cluster_optimal(gr)
plot(fc, gr,
     edge.width=10*E(gr)$weight,
     vertex.size=10*cmax,
     vertex.label.cex=0.75,
     vertex.label.dist=0,
     main="NL vs Dementia Diagnosis")

par(op)

4.0.4 NL vs Dementia table

clusterFeatures <- fc$names

tableNL_DE <- sm$coefficients[clusterFeatures,
                                   c("Estimate",
                                     "lower",
                                     "OR",
                                     "upper",
                                     "full.AUC",
                                     "Delta.AUC",
                                     "z.IDI",
                                     "Frequency")]

nugget <- fc$membership
names(nugget) <- clusterFeatures

tableNL_DE$Cluster <- nugget[rownames(tableNL_DE)]

rnames <- clusterFeatures[str_detect(clusterFeatures,"ST")]
frnames <- rnames
rnames <- str_replace_all(rnames,"M_","")
rnames <- str_replace_all(rnames,"RD_","")
description <- character()

for (ddet in c(1:length(rnames)))
{
  description <- c(description,TADPOLE_D1_D2_Dict$TEXT[str_detect(TADPOLE_D1_D2_Dict$FLDNAME,rnames[ddet])][1])
}
names(description) <- frnames

tableNL_DE$Description <- description[rownames(tableNL_DE)]
pander::pander(tableNL_DE)

	Estimate	lower	OR	upper	full.AUC	Delta.AUC	z.IDI	Frequency	Cluster	Description
M_ST24TA	-153.90	6.39e-92	1.45e-67	3.29e-43	0.915	0.02455	6.96	1.00	1	Cortical Thickness Average of LeftEntorhinal
Hippocampus	-58.83	1.17e-38	2.83e-26	6.84e-14	0.915	0.02196	5.84	1.00	1	NA
RD_ST32TA	4.64	1.49e+00	1.04e+02	7.25e+03	0.909	0.00294	3.56	0.80	1	Cortical Thickness Average of LeftInferiorTemporal
M_ST51TA	324.67	9.53e+92	1.01e+141	1.07e+189	0.916	0.03333	7.30	0.85	1	Cortical Thickness Average of LeftPrecentral
M_ST31TA	-212.64	5.58e-133	4.47e-93	3.58e-53	0.914	0.02753	6.89	0.55	1	Cortical Thickness Average of LeftInferiorParietal
M_ST24CV	-126.03	5.06e-66	1.84e-55	6.68e-45	0.896	0.04885	9.09	1.00	2	Volume (Cortical Parcellation) of LeftEntorhinal
M_ST24SA	17.83	1.47e+01	5.56e+07	2.10e+14	0.900	0.01248	4.14	0.45	2	Surface Area of LeftEntorhinal
M_ST40CV	-33.71	2.01e-25	2.29e-15	2.60e-05	0.903	0.01798	4.79	0.65	2	Volume (Cortical Parcellation) of LeftMiddleTemporal
M_ST31CV	-21.05	4.55e-16	7.24e-10	1.15e-03	0.912	0.02055	4.91	0.35	1	Volume (Cortical Parcellation) of LeftInferiorParietal
M_ST12SV	-99.35	3.37e-62	7.11e-44	1.50e-25	0.897	0.01839	5.39	0.95	2	Volume (WM Parcellation) of LeftAmygdala
M_ST15TA	-111.39	6.38e-84	4.20e-49	2.77e-14	0.911	0.01467	4.46	0.30	1	Cortical Thickness Average of LeftCaudalMiddleFrontal
M_ST32TA	-59.88	6.78e-40	9.85e-27	1.43e-13	0.889	0.01766	4.80	0.40	2	Cortical Thickness Average of LeftInferiorTemporal
RD_ST57TA	5.10	2.67e+00	1.64e+02	1.00e+04	0.897	0.01429	4.01	0.45	2	Cortical Thickness Average of LeftSuperiorParietal
M_ST36CV	41.51	5.35e+06	1.07e+18	2.12e+29	0.898	0.01678	5.12	0.55	2	Volume (Cortical Parcellation) of LeftLateralOrbitofrontal
M_ST32CV	-30.61	2.31e-23	5.09e-14	1.12e-04	0.886	0.01251	4.38	0.60	3	Volume (Cortical Parcellation) of LeftInferiorTemporal
M_ST39CV	31.34	1.34e+05	4.09e+13	1.25e+22	0.894	0.01809	5.02	0.60	2	Volume (Cortical Parcellation) of LeftMedialOrbitofrontal
M_ST30SV	9.76	6.24e+01	1.73e+04	4.79e+06	0.888	0.01592	5.42	0.40	3	Volume (WM Parcellation) of LeftInferiorLateralVentricle
M_ST40TA	-207.63	4.70e-123	6.70e-91	9.57e-59	0.888	0.02418	6.72	0.50	3	Cortical Thickness Average of LeftMiddleTemporal
RD_ST55TA	3.07	1.38e+00	2.15e+01	3.35e+02	0.902	0.01329	3.46	0.25	2	Cortical Thickness Average of LeftRostralMiddleFrontal
RD_ST29SV	3.90	2.38e+00	4.94e+01	1.02e+03	0.887	0.01862	3.99	0.25	3	Volume (WM Parcellation) of LeftHippocampus
M_ST45TA	162.32	1.76e+44	3.13e+70	5.57e+96	0.889	0.02181	5.79	0.35	3	Cortical Thickness Average of LeftParsOpercularis

4.1 Decorrelated Set

TADPOLE_DX_NLDE_TRAIND <- GDSTMDecorrelation(TADPOLE_DX_NLDE_TRAIN,Outcome="DX",
                                        thr=0.6,
                                        type="RLM",
                                        method="spearman",
                                        verbose = TRUE)
#> 
#>  Included: 194 , Uni p: 0.01165645 To Outcome: 110 , Base: 5 , In Included: 5 , Base Cor: 13 
#> 1 , Top: 38 < 0.6 >( 2 )[ 1 : 0 : 0.594 ]( 38 , 84 , 0 ),<>Tot Used: 122 , Added: 84 , Zero Std: 0 , Max Cor: 0.996129 
#> 2 , Top: 27 < 0.6 >( 1 )[ 1 : 0 : 0.594 ]( 25 , 32 , 38 ),<>Tot Used: 148 , Added: 32 , Zero Std: 0 , Max Cor: 0.9334342 
#> 3 , Top: 20 < 0.6 >[ FALSE ]( 1 )[ 1 : 1 : 0.594 ]( 19 , 25 , 59 ),<>Tot Used: 160 , Added: 25 , Zero Std: 0 , Max Cor: 0.9362059 
#> 4 , Top: 11 < 0.6 >( 3 )[ 1 : 0 : 0 ]( 11 , 14 , 72 ),<>Tot Used: 163 , Added: 14 , Zero Std: 0 , Max Cor: 0.850563 
#> 5 , Top: 5 < 0.6 >[ FALSE ]( 1 )[ 1 : 0 : 0 ]( 5 , 5 , 80 ),<>Tot Used: 165 , Added: 5 , Zero Std: 0 , Max Cor: 0.8831114 
#> 6 , Top: 3 < 0.6 >[ FALSE ]( 1 )[ 1 : 0 : 0 ]( 3 , 3 , 83 ),<>Tot Used: 167 , Added: 3 , Zero Std: 0 , Max Cor: 0.5980317 
#> [ 7 ], 0.5980317 . Cor to Base: 99 , ABase: 60
TADPOLE_DX_NLDE_TESTD <-  predictDecorrelate(TADPOLE_DX_NLDE_TRAIND,TADPOLE_DX_NLDE_TEST)

4.1.1 Learning

bDXmlNLDED <- BSWiMS.model(DX~.,TADPOLE_DX_NLDE_TRAIND,NumberofRepeats = 20)

[+-+-+-+-+-+-+-+-+–+-+-+-+-+-+–+-+-+-+-+-]..

pander::pander(bDXmlNLDED$bagging$Jaccard.SM)

1

fs <- bDXmlNLDED$bagging$frequencyTable
barplot(fs[order(-fs)],las=2,main="Selected Features",cex.names = 0.5)

sm <- summary(bDXmlNLDED)
pander::pander(sm$coefficients)

	Estimate	lower	OR	upper	u.Accuracy	r.Accuracy	full.Accuracy	u.AUC	r.AUC	full.AUC	IDI	NRI	z.IDI	z.NRI	Delta.AUC	Frequency
Ba_M_ST24TA	-719	0.00e+00	3.70e-313	4.56e-269	0.850	0.671	0.918	0.842	0.659	0.916	0.5437	1.600	26.42	31.5	0.2565	1
De_M_ST51TA	2127	Inf	Inf	Inf	0.627	0.883	0.918	0.616	0.879	0.916	0.1189	0.891	8.90	11.8	0.0361	1
De_Hippocampus	-170	2.44e-98	2.13e-74	1.86e-50	0.575	0.891	0.918	0.576	0.888	0.916	0.0783	0.783	7.24	10.1	0.0272	1

pander::pander(bDXmlNLDED$univariate[bDXmlNLDED$selectedfeatures,])

	Name	RName	ZUni
Ba_M_ST24TA	Ba_M_ST24TA	Ba_M_ST24TA	20.33
De_M_ST51TA	De_M_ST51TA	De_M_ST51TA	6.84
De_Hippocampus	De_Hippocampus	De_Hippocampus	4.23

prBin <- predictionStats_binary(cbind(TADPOLE_DX_NLDE_TESTD$DX,predict(bDXmlNLDED,TADPOLE_DX_NLDE_TESTD)),"NL vs Dementia")

NL vs Dementia

pander::pander(prBin$aucs)

est	lower	upper
0.926	0.892	0.96

pander::pander(prBin$accc)

est	lower	upper
0.867	0.822	0.905

pander::pander(prBin$berror)

50%	2.5%	97.5%
0.154	0.113	0.2

pander::pander(prBin$sensitivity)

est	lower	upper
0.765	0.669	0.845

par(op)

4.1.2 The formula network

cmax <- apply(bDXmlNLDED$bagging$formulaNetwork,2,max)
cnames <- names(cmax[cmax>=0.25])
cmax <- cmax[cnames]

adma <- bDXmlNLDED$bagging$formulaNetwork[cnames,cnames]

adma[adma<0.15] <- 0;
gr <- graph_from_adjacency_matrix(adma,mode = "undirected",diag = FALSE,weighted=TRUE)
gr$layout <- layout_with_fr

fc <- cluster_optimal(gr)
plot(fc, gr,
     edge.width=10*E(gr)$weight,
     vertex.size=10*cmax,
     vertex.label.cex=0.75,
     vertex.label.dist=0,
     main="NL vs Dementia Diagnosis")

par(op)

4.1.3 Decorrelated NL vs Dementia table

clusterFeatures <- fc$names

tableNL_DED <- sm$coefficients[clusterFeatures,
                                   c("Estimate",
                                     "lower",
                                     "OR",
                                     "upper",
                                     "full.AUC",
                                     "Delta.AUC",
                                     "z.IDI",
                                     "Frequency")]

nugget <- fc$membership
names(nugget) <- clusterFeatures

tableNL_DED$Cluster <- nugget[rownames(tableNL_DED)]

rnames <- clusterFeatures[str_detect(clusterFeatures,"ST")]
frnames <- rnames
rnames <- str_replace_all(rnames,"M_","")
rnames <- str_replace_all(rnames,"RD_","")
rnames <- str_replace_all(rnames,"Ba_","")
rnames <- str_replace_all(rnames,"De_","")
description <- character()

for (ddet in c(1:length(rnames)))
{
  description <- c(description,TADPOLE_D1_D2_Dict$TEXT[str_detect(TADPOLE_D1_D2_Dict$FLDNAME,rnames[ddet])][1])
}
names(description) <- frnames

tableNL_DED$Description <- description[rownames(tableNL_DED)]



## Getting the decorrelation formula
dc <- getDerivedCoefficients(TADPOLE_DX_NLDE_TRAIND)
decornames <- rownames(sm$coefficients)

deNames_in_dc <- decornames[decornames %in% names(dc)]
theDeFormulas <- dc[deNames_in_dc]
deFromula <- character(length(theDeFormulas))
names(deFromula) <- names(theDeFormulas)
for (dx in names(deFromula))
{
  coef <- theDeFormulas[[dx]]
  cname <- names(theDeFormulas[[dx]])
  names(cname) <- cname
  for (cf in names(coef))
  {
    if (cf != dx)
    {
      if (coef[cf]>0)
      {
        deFromula[dx] <- paste(deFromula[dx],
                               sprintf("+ %5.3f*%s",coef[cf],cname[cf]))
      }
      else
      {
        deFromula[dx] <- paste(deFromula[dx],
                               sprintf("%5.3f*%s",coef[cf],cname[cf]))
      }
    }
  }
}

tableNL_DED$DecorFormula <- deFromula[rownames(tableNL_DED)]



pander::pander(tableNL_DED)

	Estimate	lower	OR	upper	full.AUC	Delta.AUC	z.IDI	Frequency	Cluster	Description	DecorFormula
Ba_M_ST24TA	-719	0.00e+00	3.70e-313	4.56e-269	0.916	0.2565	26.42	1	1	Cortical Thickness Average of LeftEntorhinal	NA
De_M_ST51TA	2127	Inf	Inf	Inf	0.916	0.0361	8.90	1	1	Cortical Thickness Average of LeftPrecentral	-0.515M_ST31TA + 1.000M_ST51TA + 0.032M_ST51SA -0.135M_ST51CV
De_Hippocampus	-170	2.44e-98	2.13e-74	1.86e-50	0.916	0.0272	7.24	1	1	NA	+ 1.000Hippocampus -1.820M_ST24TA

5 Prognosis MCI to AD Conversion

5.1 the set

TrainFraction <- 0.60;

MCIPrognosisIDs <- c(MCIconverters$PTID,MCI_No_converters$PTID)

TADPOLECrossMRI <- validBaselineTadpole[MCIPrognosisIDs,]
table(TADPOLECrossMRI$DX)

Dementia MCI NL 0 680 0

TADPOLECrossMRI$DX <- NULL
TADPOLECrossMRI$status <- 1*(rownames(TADPOLECrossMRI) %in% MCIconverters$PTID)
table(TADPOLECrossMRI$status)

0 1 436 244

TADPOLECrossMRI$TimeToEvent <- numeric(nrow(TADPOLECrossMRI))
TADPOLECrossMRI[MCIconverters$PTID,"TimeToEvent"] <- MCIconverters$TimeToEvent
TADPOLECrossMRI[MCI_No_converters$PTID,"TimeToEvent"] <- MCI_No_converters$TimeToEvent

TADPOLE_Cases <- subset(TADPOLECrossMRI,status==1)
TADPOLE_Controls <- subset(TADPOLECrossMRI,status==0)
trainCasesSet <- sample(nrow(TADPOLE_Cases),nrow(TADPOLE_Cases)*TrainFraction)
trainControlSet <- sample(nrow(TADPOLE_Controls),nrow(TADPOLE_Controls)*TrainFraction)

TADPOLE_Conv_TRAIN <- rbind(TADPOLE_Cases[trainCasesSet,],TADPOLE_Controls[trainControlSet,])
TADPOLE_Conv_TEST <- TADPOLECrossMRI[!(rownames(TADPOLECrossMRI) %in%
                                         rownames(TADPOLE_Conv_TRAIN)),]

pander::pander(table(TADPOLE_Conv_TRAIN$status))

0	1
261	146

pander::pander(table(TADPOLE_Conv_TEST$status))

0	1
175	98

par(op)

5.1.1 Learning

bConvml <- BSWiMS.model(Surv(TimeToEvent,status)~1,TADPOLE_Conv_TRAIN,NumberofRepeats = 20)

[++–++–+-+-++-++-+++-++-+-+++-++-++-++-++–+++-+–+++—+–+–++-+-+-+++-]….

pander::pander(bConvml$bagging$Jaccard.SM)

0.257

fs <- bConvml$bagging$frequencyTable
barplot(fs[order(-fs)],las=2,main="Selected Features",cex.names = 0.5)

sm <- summary(bConvml)
pander::pander(sm$coefficients)

	Estimate	lower	HR	upper	u.Accuracy	r.Accuracy	full.Accuracy	u.AUC	r.AUC	full.AUC	IDI	NRI	z.IDI	z.NRI	Delta.AUC	Frequency
ADAS13	0.06013	1.05e+00	1.06e+00	1.08e+00	0.713	0.759	0.781	0.712	0.750	0.781	0.0853	0.635	8.67	7.69	0.030593	1.00
M_ST24CV	-15.56540	4.49e-09	1.74e-07	6.72e-06	0.692	0.684	0.735	0.680	0.698	0.742	0.0690	0.657	8.14	8.01	0.044068	0.35
RAVLT_immediate	-0.01969	9.75e-01	9.81e-01	9.86e-01	0.633	0.725	0.736	0.647	0.719	0.744	0.0895	0.624	7.78	7.61	0.025171	0.50
ADAS11	0.02288	1.01e+00	1.02e+00	1.03e+00	0.687	0.706	0.747	0.674	0.710	0.749	0.0767	0.600	7.47	7.28	0.038270	0.75
M_ST40SA	-3.62353	5.48e-03	2.67e-02	1.30e-01	0.622	0.732	0.734	0.621	0.740	0.741	0.0336	0.578	6.43	6.97	0.001001	0.40
M_ST12SV	-12.77498	1.46e-08	2.83e-06	5.48e-04	0.695	0.731	0.748	0.697	0.727	0.751	0.0394	0.540	6.19	6.47	0.023798	0.75
M_ST40CV	-16.30048	3.77e-12	8.33e-08	1.84e-03	0.704	0.763	0.781	0.692	0.759	0.781	0.0277	0.381	5.20	4.47	0.022549	1.00
RAVLT_perc_forgetting	0.00232	1.00e+00	1.00e+00	1.00e+00	0.671	0.777	0.779	0.682	0.775	0.779	0.0150	0.601	4.96	7.32	0.004889	0.90
FAQ	0.05881	1.02e+00	1.06e+00	1.10e+00	0.735	0.758	0.781	0.696	0.761	0.781	0.0315	0.492	4.78	6.10	0.020034	1.00
RAVLT_forgetting	0.01450	1.00e+00	1.01e+00	1.03e+00	0.563	0.737	0.736	0.578	0.744	0.744	0.0134	0.447	4.74	5.28	0.000192	0.30
M_ST56CV	-6.06725	2.77e-05	2.32e-03	1.94e-01	0.656	0.759	0.765	0.652	0.755	0.766	0.0230	0.344	4.64	4.01	0.010988	0.50
M_ST24SA	-9.47742	3.19e-08	7.66e-05	1.84e-01	0.595	0.777	0.781	0.596	0.775	0.782	0.0200	0.400	4.24	4.68	0.006677	0.85
RD_ST65SV	-6.62059	4.08e-06	1.33e-03	4.35e-01	0.530	0.719	0.739	0.553	0.725	0.745	0.0192	0.359	4.18	4.24	0.020334	0.50
RD_ST56TA	2.57234	1.68e+00	1.31e+01	1.02e+02	0.604	0.738	0.736	0.574	0.751	0.743	0.0183	0.295	3.97	3.51	-0.007372	0.25
RD_ST31TA	1.11643	1.15e+00	3.05e+00	8.09e+00	0.628	0.764	0.771	0.605	0.763	0.769	0.0134	0.343	3.54	4.05	0.006803	0.55

pander::pander(bConvml$univariate[bConvml$selectedfeatures,])

	Name	RName	ZUni
ADAS13	ADAS13	ADAS13	11.01
FAQ	FAQ	FAQ	8.43
M_ST40CV	M_ST40CV	M_ST40CV	10.28
M_ST24SA	M_ST24SA	M_ST24SA	3.84
RAVLT_perc_forgetting	RAVLT_perc_forgetting	RAVLT_perc_forgetting	7.98
ADAS11	ADAS11	ADAS11	9.67
M_ST29SV	M_ST29SV	M_ST29SV	9.26
RD_ST31TA	RD_ST31TA	RD_ST31TA	4.27
M_ST12SV	M_ST12SV	M_ST12SV	9.67
M_ST56CV	M_ST56CV	M_ST56CV	7.58
RAVLT_immediate	RAVLT_immediate	RAVLT_immediate	7.50
M_ST40SA	M_ST40SA	M_ST40SA	5.61
RD_ST65SV	RD_ST65SV	RD_ST65SV	1.74
RAVLT_forgetting	RAVLT_forgetting	RAVLT_forgetting	3.47
M_ST24CV	M_ST24CV	M_ST24CV	8.21
M_ST13TA	M_ST13TA	M_ST13TA	8.07
RD_ST56TA	RD_ST56TA	RD_ST56TA	3.22
RD_ST49CV	RD_ST49CV	RD_ST49CV	1.74

ptestl <- predict(bConvml,TADPOLE_Conv_TEST,type="lp")
boxplot(ptestl~TADPOLE_Conv_TEST$status)

ptestr <- predict(bConvml,TADPOLE_Conv_TEST,type="risk")
eventCases <- subset(TADPOLE_Conv_TEST,status==1)
plot(1.0/ptestr[rownames(eventCases)]~eventCases$TimeToEvent)

pander::pander(cor.test(eventCases$TimeToEvent,1.0/ptestr[rownames(eventCases)],method="spearman"))

Spearman’s rank correlation rho: eventCases$TimeToEvent and 1/ptestr[rownames(eventCases)]

Test statistic	P value	Alternative hypothesis	rho
110840	0.00337 * *	two.sided	0.293

perdsurv <- cbind(TADPOLE_Conv_TEST$TimeToEvent,
                  TADPOLE_Conv_TEST$status,
                  ptestl,
                  ptestr)
prSurv <- predictionStats_survival(perdsurv,"MCI to  AD Conversion")

pander::pander(prSurv$CIRisk)

median	lower	upper
0.829	0.793	0.865

pander::pander(prSurv$CILp)

median	lower	upper
0.86	0.813	0.903

pander::pander(prSurv$spearmanCI)

50%	2.5%	97.5%
0.295	0.0855	0.488

prBin <- predictionStats_binary(cbind(TADPOLE_Conv_TEST$status,ptestl),"MCI to  AD Conversion")

MCI to AD Conversion

pander::pander(prBin$aucs)

est	lower	upper
0.86	0.816	0.904

pander::pander(prBin$CM.analysis$tab)

	Outcome +	Outcome -	Total
Test +	78	38	116
Test -	20	137	157
Total	98	175	273

par(op)

5.1.2 The formula network

cmax <- apply(bConvml$bagging$formulaNetwork,2,max)
cnames <- names(cmax[cmax>=0.25])
cmax <- cmax[cnames]

adma <- bConvml$bagging$formulaNetwork[cnames,cnames]

adma[adma<0.15] <- 0;
gr <- graph_from_adjacency_matrix(adma,mode = "undirected",diag = FALSE,weighted=TRUE)
gr$layout <- layout_with_fr

fc <- cluster_optimal(gr)
plot(fc, gr,
     edge.width=5*E(gr)$weight,
     vertex.size=20*cmax,
     vertex.label.cex=0.75,
     vertex.label.dist=0,
     main="MCI to Dementia Conversion")

par(op)

5.1.3 MCI to Dementia table

clusterFeatures <- fc$names

tableMCI_to_Dem <- sm$coefficients[clusterFeatures,
                                   c("Estimate",
                                     "lower",
                                     "HR",
                                     "upper",
                                     "full.AUC",
                                     "Delta.AUC",
                                     "z.IDI",
                                     "Frequency")]

nugget <- fc$membership
names(nugget) <- clusterFeatures

tableMCI_to_Dem$Cluster <- nugget[rownames(tableMCI_to_Dem)]

rnames <- clusterFeatures[str_detect(clusterFeatures,"ST")]
frnames <- rnames
rnames <- str_replace_all(rnames,"M_","")
rnames <- str_replace_all(rnames,"RD_","")
rnames <- str_replace_all(rnames,"Ba_","")
rnames <- str_replace_all(rnames,"De_","")
description <- character()

for (ddet in c(1:length(rnames)))
{
  description <- c(description,TADPOLE_D1_D2_Dict$TEXT[str_detect(TADPOLE_D1_D2_Dict$FLDNAME,rnames[ddet])][1])
}
names(description) <- frnames

tableMCI_to_Dem$Description <- description[rownames(tableMCI_to_Dem)]
pander::pander(tableMCI_to_Dem)

	Estimate	lower	HR	upper	full.AUC	Delta.AUC	z.IDI	Frequency	Cluster	Description
ADAS13	0.06013	1.05e+00	1.06e+00	1.08e+00	0.781	0.030593	8.67	1.00	1	NA
FAQ	0.05881	1.02e+00	1.06e+00	1.10e+00	0.781	0.020034	4.78	1.00	1	NA
M_ST40CV	-16.30048	3.77e-12	8.33e-08	1.84e-03	0.781	0.022549	5.20	1.00	1	Volume (Cortical Parcellation) of LeftMiddleTemporal
M_ST24SA	-9.47742	3.19e-08	7.66e-05	1.84e-01	0.782	0.006677	4.24	0.85	1	Surface Area of LeftEntorhinal
RAVLT_perc_forgetting	0.00232	1.00e+00	1.00e+00	1.00e+00	0.779	0.004889	4.96	0.90	1	NA
ADAS11	0.02288	1.01e+00	1.02e+00	1.03e+00	0.749	0.038270	7.47	0.75	2	NA
RD_ST31TA	1.11643	1.15e+00	3.05e+00	8.09e+00	0.769	0.006803	3.54	0.55	2	Cortical Thickness Average of LeftInferiorParietal
M_ST12SV	-12.77498	1.46e-08	2.83e-06	5.48e-04	0.751	0.023798	6.19	0.75	2	Volume (WM Parcellation) of LeftAmygdala
M_ST56CV	-6.06725	2.77e-05	2.32e-03	1.94e-01	0.766	0.010988	4.64	0.50	2	Volume (Cortical Parcellation) of LeftSuperiorFrontal
RAVLT_immediate	-0.01969	9.75e-01	9.81e-01	9.86e-01	0.744	0.025171	7.78	0.50	3	NA
M_ST40SA	-3.62353	5.48e-03	2.67e-02	1.30e-01	0.741	0.001001	6.43	0.40	3	Surface Area of LeftMiddleTemporal
RD_ST65SV	-6.62059	4.08e-06	1.33e-03	4.35e-01	0.745	0.020334	4.18	0.50	3	Volume (WM Parcellation) of LeftVentralDC
RAVLT_forgetting	0.01450	1.00e+00	1.01e+00	1.03e+00	0.744	0.000192	4.74	0.30	3	NA
M_ST24CV	-15.56540	4.49e-09	1.74e-07	6.72e-06	0.742	0.044068	8.14	0.35	3	Volume (Cortical Parcellation) of LeftEntorhinal
RD_ST56TA	2.57234	1.68e+00	1.31e+01	1.02e+02	0.743	-0.007372	3.97	0.25	3	Cortical Thickness Average of LeftSuperiorFrontal

5.2 Decorrelated

TADPOLE_Conv_TRAIND <- GDSTMDecorrelation(TADPOLE_Conv_TRAIN,
                                          Outcome="status",
                                          thr=0.6,
                                          type="RLM",
                                          method="spearman",
                                          verbose = TRUE)
#> 
#>  Included: 212 , Uni p: 0.012727 To Outcome: 119 , Base: 4 , In Included: 4 , Base Cor: 6 
#> 1 , Top: 40 < 0.6 >( 3 )[ 1 : 0 : 0.594 ]( 38 , 100 , 0 ),<>Tot Used: 138 , Added: 100 , Zero Std: 0 , Max Cor: 0.9930238 
#> 2 , Top: 36 < 0.6 >[ FALSE ]( 1 )[ 1 : 0 : 0.594 ]( 33 , 40 , 38 ),<>Tot Used: 168 , Added: 40 , Zero Std: 0 , Max Cor: 0.8953444 
#> 3 , Top: 22 < 0.6 >( 1 )[ 1 : 0 : 0.594 ]( 22 , 25 , 64 ),<>Tot Used: 171 , Added: 25 , Zero Std: 0 , Max Cor: 0.9205408 
#> 4 , Top: 10 < 0.6 >( 1 )[ 1 : 0 : 0 ]( 10 , 11 , 81 ),<>Tot Used: 172 , Added: 11 , Zero Std: 0 , Max Cor: 0.6303874 
#> 5 , Top: 2 < 0.6 >( 1 )[ 1 : 0 : 0.6 ]( 2 , 2 , 87 ),<>Tot Used: 172 , Added: 2 , Zero Std: 0 , Max Cor: 0.5997743 
#> [ 6 ], 0.5997743 . Cor to Base: 107 , ABase: 77
TADPOLE_Conv_TESTD <-  predictDecorrelate(TADPOLE_Conv_TRAIND,TADPOLE_Conv_TEST)

5.2.1 Learning

bConvmlD <- BSWiMS.model(Surv(TimeToEvent,status)~1,TADPOLE_Conv_TRAIND,NumberofRepeats = 20)

[+-+-+-+-+-+-+-+-+–+-+-+-+-+-+-+-+-+–+-+-]..

pander::pander(bConvmlD$bagging$Jaccard.SM)

0.617

fs <- bConvmlD$bagging$frequencyTable
barplot(fs[order(-fs)],las=2,main="Selected Features",cex.names = 0.5)

sm <- summary(bConvmlD)
pander::pander(sm$coefficients)

	Estimate	lower	HR	upper	u.Accuracy	r.Accuracy	full.Accuracy	u.AUC	r.AUC	full.AUC	IDI	NRI	z.IDI	z.NRI	Delta.AUC	Frequency
Ba_ADAS13	0.1173	1.10e+00	1.12e+00	1.15e+00	0.713	0.773	0.796	0.712	0.761	0.795	0.108962	0.698	9.66	8.57	0.03406	1.00
FAQ	0.1242	1.08e+00	1.13e+00	1.19e+00	0.734	0.768	0.796	0.696	0.770	0.795	0.050598	0.571	6.63	7.23	0.02410	1.00
De_M_ST13TA	-258.3195	1.34e-182	6.51e-113	3.15e-43	0.599	0.782	0.798	0.598	0.777	0.797	0.028197	0.445	5.23	5.25	0.01923	0.75
M_ST49CV	-26.8013	4.50e-20	2.29e-12	1.17e-04	0.606	0.796	0.797	0.613	0.790	0.795	0.019705	0.409	4.92	4.80	0.00478	0.75
RD_ST31TA	5.9872	6.48e+00	3.98e+02	2.45e+04	0.627	0.796	0.798	0.605	0.793	0.797	0.017701	0.471	4.82	5.64	0.00397	0.80
De_M_ST48CV	-8.8309	3.52e-07	1.46e-04	6.06e-02	0.564	0.809	0.814	0.554	0.808	0.814	0.007397	0.264	4.14	3.06	0.00624	0.25
Ba_RD_ST49CV	-0.6003	3.32e-01	5.49e-01	9.07e-01	0.519	0.818	0.828	0.532	0.818	0.828	0.013152	0.267	4.09	3.12	0.00956	0.10
De_M_ST24SA	-16.1212	8.69e-14	9.97e-08	1.14e-01	0.592	0.786	0.797	0.597	0.781	0.795	0.014967	0.362	3.91	4.23	0.01388	0.90
De_M_ST29SV	-3.2129	3.61e-03	4.02e-02	4.49e-01	0.706	0.791	0.799	0.698	0.786	0.796	0.010221	0.259	3.76	3.01	0.00980	0.35
RD_ST37SV	-0.5303	3.46e-01	5.88e-01	1.00e+00	0.495	0.799	0.802	0.521	0.799	0.802	0.010429	0.293	3.49	3.45	0.00306	0.10
RD_ST65SV	-4.9615	5.09e-05	7.00e-03	9.64e-01	0.531	0.798	0.810	0.552	0.798	0.810	0.007927	0.297	3.24	3.49	0.01159	0.30
Ba_M_ST56SA	-1.3203	8.06e-02	2.67e-01	8.85e-01	0.566	0.798	0.820	0.565	0.798	0.822	0.014117	0.171	3.24	1.98	0.02373	0.15
De_M_ST39CV	-0.0835	7.29e-01	9.20e-01	1.16e+00	0.568	0.780	0.783	0.563	0.770	0.774	0.000919	0.196	2.09	2.26	0.00429	0.05

pander::pander(bConvmlD$univariate[bConvmlD$selectedfeatures,])

	Name	RName	ZUni
Ba_ADAS13	Ba_ADAS13	Ba_ADAS13	11.01
FAQ	FAQ	FAQ	8.43
De_M_ST29SV	De_M_ST29SV	De_M_ST29SV	8.93
M_ST49CV	M_ST49CV	M_ST49CV	5.76
De_M_ST39CV	De_M_ST39CV	De_M_ST39CV	3.14
De_M_ST24SA	De_M_ST24SA	De_M_ST24SA	3.84
De_M_ST13TA	De_M_ST13TA	De_M_ST13TA	4.26
RD_ST31TA	RD_ST31TA	RD_ST31TA	4.27
De_M_ST48CV	De_M_ST48CV	De_M_ST48CV	3.38
RD_ST37SV	RD_ST37SV	RD_ST37SV	1.66
RD_ST65SV	RD_ST65SV	RD_ST65SV	1.74
Ba_M_ST56SA	Ba_M_ST56SA	Ba_M_ST56SA	3.97
Ba_RD_ST49CV	Ba_RD_ST49CV	Ba_RD_ST49CV	1.74

ptestl <- predict(bConvmlD,TADPOLE_Conv_TESTD,type="lp")
boxplot(ptestl~TADPOLE_Conv_TEST$status)

ptestr <- predict(bConvmlD,TADPOLE_Conv_TESTD,type="risk")
eventCases <- subset(TADPOLE_Conv_TEST,status==1)
plot(1.0/ptestr[rownames(eventCases)]~eventCases$TimeToEvent)

pander::pander(cor.test(eventCases$TimeToEvent,1.0/ptestr[rownames(eventCases)],method="spearman"))

Spearman’s rank correlation rho: eventCases$TimeToEvent and 1/ptestr[rownames(eventCases)]

Test statistic	P value	Alternative hypothesis	rho
103978	0.000689 * * *	two.sided	0.337

perdsurv <- cbind(TADPOLE_Conv_TEST$TimeToEvent,
                  TADPOLE_Conv_TEST$status,
                  ptestl,
                  ptestr)
prSurv <- predictionStats_survival(perdsurv,"MCI to  AD Conversion")

pander::pander(prSurv$CIRisk)

median	lower	upper
0.83	0.791	0.866

pander::pander(prSurv$CILp)

median	lower	upper
0.857	0.807	0.899

pander::pander(prSurv$spearmanCI)

50%	2.5%	97.5%
0.34	0.134	0.521

prBin <- predictionStats_binary(cbind(TADPOLE_Conv_TESTD$status,ptestl),"MCI to  AD Conversion")

MCI to AD Conversion

pander::pander(prBin$aucs)

est	lower	upper
0.857	0.813	0.902

pander::pander(prBin$CM.analysis$tab)

	Outcome +	Outcome -	Total
Test +	76	36	112
Test -	22	139	161
Total	98	175	273

par(op)

5.2.2 The formula network

cmax <- apply(bConvmlD$bagging$formulaNetwork,2,max)
cnames <- names(cmax[cmax>=0.25])
cmax <- cmax[cnames]

adma <- bConvmlD$bagging$formulaNetwork[cnames,cnames]

adma[adma<0.15] <- 0;
gr <- graph_from_adjacency_matrix(adma,mode = "undirected",diag = FALSE,weighted=TRUE)
gr$layout <- layout_with_fr

fc <- cluster_optimal(gr)
plot(fc, gr,
     edge.width=5*E(gr)$weight,
     vertex.size=20*cmax,
     vertex.label.cex=0.75,
     vertex.label.dist=0,
     main="MCI to Dementia Conversion")

par(op)

5.2.3 Decorrelated MCI to Dementia table

clusterFeatures <- fc$names

tableMCI_to_DemD <- sm$coefficients[clusterFeatures,
                                   c("Estimate",
                                     "lower",
                                     "HR",
                                     "upper",
                                     "full.AUC",
                                     "Delta.AUC",
                                     "z.IDI",
                                     "Frequency")]

nugget <- fc$membership
names(nugget) <- clusterFeatures

tableMCI_to_DemD$Cluster <- nugget[rownames(tableMCI_to_DemD)]

rnames <- clusterFeatures[str_detect(clusterFeatures,"ST")]
frnames <- rnames
rnames <- str_replace_all(rnames,"M_","")
rnames <- str_replace_all(rnames,"RD_","")
rnames <- str_replace_all(rnames,"Ba_","")
rnames <- str_replace_all(rnames,"De_","")
description <- character()

for (ddet in c(1:length(rnames)))
{
  description <- c(description,TADPOLE_D1_D2_Dict$TEXT[str_detect(TADPOLE_D1_D2_Dict$FLDNAME,rnames[ddet])][1])
}
names(description) <- frnames

tableMCI_to_DemD$Description <- description[rownames(tableMCI_to_DemD)]


## Getting the decorrelation formula
dc <- getDerivedCoefficients(TADPOLE_Conv_TRAIND)
decornames <- rownames(sm$coefficients)

deNames_in_dc <- decornames[decornames %in% names(dc)]
theDeFormulas <- dc[deNames_in_dc]
deFromula <- character(length(theDeFormulas))
names(deFromula) <- names(theDeFormulas)
for (dx in names(deFromula))
{
  coef <- theDeFormulas[[dx]]
  cname <- names(theDeFormulas[[dx]])
  names(cname) <- cname
  for (cf in names(coef))
  {
    if (cf != dx)
    {
      if (coef[cf]>0)
      {
        deFromula[dx] <- paste(deFromula[dx],
                               sprintf("+ %5.3f*%s",coef[cf],cname[cf]))
      }
      else
      {
        deFromula[dx] <- paste(deFromula[dx],
                               sprintf("%5.3f*%s",coef[cf],cname[cf]))
      }
    }
  }
}

tableMCI_to_DemD$DecorFormula <- deFromula[rownames(tableMCI_to_DemD)]


pander::pander(tableMCI_to_DemD)

	Estimate	lower	HR	upper	full.AUC	Delta.AUC	z.IDI	Frequency	Cluster	Description	DecorFormula
Ba_ADAS13	0.117	1.10e+00	1.12e+00	1.15e+00	0.795	0.03406	9.66	1.00	1	NA	NA
FAQ	0.124	1.08e+00	1.13e+00	1.19e+00	0.795	0.02410	6.63	1.00	1	NA	NA
De_M_ST29SV	-3.213	3.61e-03	4.02e-02	4.49e-01	0.796	0.00980	3.76	0.35	2	Volume (WM Parcellation) of LeftHippocampus	+ 0.105Ventricles + 1.000M_ST29SV
M_ST49CV	-26.801	4.50e-20	2.29e-12	1.17e-04	0.795	0.00478	4.92	0.75	1	Volume (Cortical Parcellation) of LeftPostcentral	NA
De_M_ST24SA	-16.121	8.69e-14	9.97e-08	1.14e-01	0.795	0.01388	3.91	0.90	1	Surface Area of LeftEntorhinal	NA
De_M_ST13TA	-258.320	1.34e-182	6.51e-113	3.15e-43	0.797	0.01923	5.23	0.75	1	Cortical Thickness Average of LeftBankssts	+ 1.000M_ST13TA -0.861M_ST59TA
RD_ST31TA	5.987	6.48e+00	3.98e+02	2.45e+04	0.797	0.00397	4.82	0.80	1	Cortical Thickness Average of LeftInferiorParietal	NA
De_M_ST48CV	-8.831	3.52e-07	1.46e-04	6.06e-02	0.814	0.00624	4.14	0.25	2	Volume (Cortical Parcellation) of LeftPericalcarine	-6.760M_ST48TS -0.203M_ST48SA + 1.000*M_ST48CV
RD_ST65SV	-4.962	5.09e-05	7.00e-03	9.64e-01	0.810	0.01159	3.24	0.30	1	Volume (WM Parcellation) of LeftVentralDC	NA

5.2.4 Saving the enviroment

save.image("~/GitHub/BSWiMS/TADPOLE_BSWIMS_Results.RData")